Citation
Optical sensors development for NDE applications using structured light and 3D visualization

Material Information

Title:
Optical sensors development for NDE applications using structured light and 3D visualization
Creator:
Elfaid, Abdusamea ( author )
Language:
English
Physical Description:
1 electronic file (78 pages). : ;

Subjects

Subjects / Keywords:
Nondestructive testing ( lcsh )
Three-dimensional imaging ( lcsh )
Nondestructive testing ( fast )
Three-dimensional imaging ( fast )
Genre:
non-fiction ( marcgt )

Notes

Review:
As pipeline infrastructure systems continue to age and deteriorate, efficient and effective repair or replacement, and maintenance scheduling to reduce the associated significant costs are critical but remain challenging. Pipeline inspection technologies and innovative solutions need to be improved and/or developed in order to provide cost effective solutions to assist the pipeline safety and reliability full decision support system. Moreover, identification and classification of current vintage pipeline inner wall damage precursor are of critical importance. However, there are limited success in sensing and characterizing the small diameters of pipelines with high probability of detection (POD) [3], and the capability of the currently available accelerometers and imaging technologies that can be miniaturized and integrated into smaller size pipes for a fast scan is questionable and needs a systematic assessment. The major drivers of premature failure due to slow crack growth are bending stresses due to tight bend radii, impingement and fittings. Damage is also introduced by pipe squeeze-off during maintenance operations. These conditions identification has been investigated by various nondestructive evaluation (NDE) [2] techniques, such as direct visual/optical methods using CCD cameras, ultrasonic testing, liquid-coupled acoustic measurement (e.g. sonar) and laser based surface inspection approaches including light detection and ranging (LiDAR)[18][19] and laser topography. However, these currently available technologies suffer either from low-sensitivity and resolution for small damage precursors, or complex settings and large system footprint that makes it incapable for plastic pipes with much smaller diameters.
Review:
In this thesis three different generations have been designed in order to obtain 3D visualization. The first generation concentrated to scanning the 3 inches pipes with linear defects. Laser source and camera with small view angle have been used to collect the data. There are missing parts in the final 3D image because the angle of the camera is small to capture all the scene. For this reason, the second generation is illustrated to decrease the shadow area that occurs from the view angle of the camera. Another camera with large view angle (fisheye camera) is used with the laser source to obtain full ring without missing parts. The third generation is structured light scanning. A new slide projector with two rings two colors has been designed to capture the scene in order to generate 3D reconstructed image. A pipe with squeezed part is used to do this test by using the structured light scanning. 3D reconstructed image illustrated the squeezed part in the pipe, and this twisted part has a shape of an ellipse instead of a circle. Then, the semi-major and semi-minor axes are calculated to define the eccentricity in order to clarify the frames with squeezed part. Misalignment correction is applied to reduce the camera tilting that occurs from using human hands while doing the scan.
Thesis:
Thesis (M.S.)--University of Colorado Denver.
Bibliography:
Includes bibliographic references.
System Details:
System requirements: Internet connectivity.
Statement of Responsibility:
by Abdusamea Elfaid.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
957592187 ( OCLC )
ocn957592187

Downloads

This item has the following downloads:


Full Text
OPTICAL SENSORS DEVELOPMENT FORNDE APPLICATIONS USING STRUCTURED
LIGHT AND 3D VISUALIZATION
By
ABDUSAMEA ELF AID
B.S., Aljabel Algharbi University, 2010
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Science
Electrical Engineering
2016


2016
ABDUSAMEA ELF AID
ALL RIGHTS RESERVED
11


This thesis for the Master of Science degree by
Abdusamea Elfaid
has been approved for the
Electrical Engineering Program
By
Yiming Deng, Chair
Jan Bialasiewicz
Chao Liu
April 28, 2016
m


Elfaid, Abdusamea. (M.S., Electrical Engineering)
Optical Sensors Development for NDE Applications Using Structured Light and 3D
Accelerating Visualization.
Thesis directed by Assistant Professor. Yiming Deng
ABSTRACT
As pipeline infrastructure systems continue to age and deteriorate, efficient and effective
repair or replacement, and maintenance scheduling to reduce the associated significant
costs are critical but remain challenging. Pipeline inspection technologies and innovative
solutions need to be improved and/or developed in order to provide cost effective solutions
to assist the pipeline safety and reliability full decision support system. Moreover,
identification and classification of current vintage pipeline inner wall damage precursor are
of critical importance. However, there are limited success in sensing and characterizing the
small diameters of pipelines with high probability of detection (POD) [3], and the
capability of the currently available accelerometers and imaging technologies that can be
miniaturized and integrated into smaller size pipes for a fast scan is questionable and needs
a systematic assessment. The major drivers of premature failure due to slow crack growth
are bending stresses due to tight bend radii, impingement and fittings. Damage is also
introduced by pipe squeeze-off during maintenance operations. These conditions
identification has been investigated by various nondestructive evaluation (NDE) [2]
techniques, such as direct visual/optical methods using CCD cameras, ultrasonic testing,
liquid-coupled acoustic measurement (e.g. sonar) and laser based surface inspection
approaches including light detection and ranging (LiDAR)[18][19] and laser topography.
However, these currently available technologies suffer either from low-sensitivity and
IV


resolution for small damage precursors, or complex settings and large system footprint that
makes it incapable for plastic pipes with much smaller diameters.
In this thesis three different generations have been designed in order to obtain 3D
visualization. The first generation concentrated to scanning the 3 inches pipes with linear
defects. Laser source and camera with small view angle have been used to collect the data.
There are missing parts in the final 3D image because the angle of the camera is small to
capture all the scene. For this reason, the second generation is illustrated to decrease the
shadow area that occurs from the view angle of the camera. Another camera with large
view angle (fisheye camera) is used with the laser source to obtain full ring without missing
parts. The third generation is structured light scanning. A new slide projector with two
rings two colors has been designed to capture the scene in order to generate 3D
reconstructed image. A pipe with squeezed part is used to do this test by using the
structured light scanning. 3D reconstructed image illustrated the squeezed part in the pipe,
and this twisted part has a shape of an ellipse instead of a circle. Then, the semi-major and
semi-minor axes are calculated to define the eccentricity in order to clarify the frames with
squeezed part. Misalignment correction is applied to reduce the camera tilting that occurs
from using human hands while doing the scan.
The form and content of this abstract are approved. I recommend its publication.
Approved by: Yiming Deng
v


ACKNOWLEDGEMENTS
I appreciate the inspiration, guidance, and assistance of many people without whom this
work would not have been possible. Professor Yiming Deng, my thesis advisor, provided
guidance and support not only for this thesis, but also in my professional development.
I would like to thank the Libyan government, especially the Ministry of Higher Education
and CBIE for nominating me to study aboard and supporting me and my family with a
fully-funded scholarship.
Finally, I must take the opportunity to acknowledge and thank my parents for their endless
love and support. They have implanted in me high work ethics and confidence to prosper.
vi


TABLE OF CONTENTS
Chapter
1. Introduction.........................................................................1
1.1 Background......................................................................1
1.2 Research Significance...........................................................2
1.3 Thesis Outline..................................................................3
2. Nondestructive Evaluation (NDE).....................................................5
2.1 Introduction....................................................................5
2.2 NDE Methods.....................................................................7
2.2.1 Visual and Optical Testing (VT)............................................7
2.2.2 Radiographic Testing (RT).....................................................8
2.2.3 Electromagnetic Testing (ET)...............................................9
2.2.4 Ultrasonic Testing (UT)...................................................11
2.2.5 Liquid penetrant Testing (PT).............................................12
2.2.6 Magnetic Particle Testing (MT)............................................14
2.2.7 Acoustic Emission Testing (AE)............................................15
2.2.8 Infrared and Thermal Testing (IR).........................................15
3. Optical Sensors....................................................................17
3.1 Introduction...................................................................17
3.2 Passive Acquisition............................................................17
3.2.1 Multiple Viewpoints.......................................................18
3.2.2 Single Viewpoints.........................................................19
3.3 Active Acquisition.............................................................20
3.3.1 Active-Stereo.............................................................20
vii


3.3.1.1 Laser Scanning
21
3.3.1.2 Structured Light......................................................22
3.3.2 Triangulation..............................................................26
3.3.3 Time-of-Flight.............................................................29
3.3.3.1 Direct Time-of-Flight.................................................29
3.3.3.1.1 LiDAR.............................................................29
3.3.3.2 Range-Gated Imaging...................................................30
3.33.3 Phase Difference Measurement..........................................31
4. 3D Pipe Inner Wall Reconstruction...................................................33
4.1. General Overview................................................................33
4.2 Materials and Preparation.......................................................33
4.2.1 The Defects and Cracks in the Pipes........................................33
4.2.2 Scanning without Light Source..............................................35
4.3. Single Ring Scanning............................................................37
4.3.1 Scanning with Small View Angle Camera......................................38
4.3.2 Scanning with Large View Angle Camera......................................41
4.3.3 Scanner and the Sensor in Opposite Direction...............................44
4.4 Multi-Rings Multi-Colors Scanning...............................................45
4.4.1 Projector Prototype...........................................................46
5. Discussion and Analysis.............................................................52
5.1 Deformation Detection.........................................................52
5.1.1 Eccentricity Calculation...................................................52
5.1.1.1 Eccentricity of the Area with Free Crack Deformation..................54
5.1.1.2 Eccentricity of the Area with Squeezed Parts..........................55
viii


5.2 Misalignment Correction....................................................58
6. Conclusion and Future Work.....................................................62
6.1 Conclusion.................................................................62
6.2 Future Work...............................................................64
References........................................................................65
IX


LIST OF FIGURES
Figure
1.1a- Illustration of ductile failure of PE pipe b-Typical slit type fractures [21].......2
2.1 sample that shows the destructive evaluation will affect the characteristic of the
spacemen [4]................................................................................5
2.2 Illustration of the old technique of using visual testing [5]..........................8
2.3 discontinuity detection by radiographic testing (x-ray) [5]............................9
2.4 The principle of electromagnetic testing [6].........................................10
2.5 Illustration of using ferrous inclusion to detect the metal [6].......................10
2.6 Illustration of main steps of Ultrasonic Testing [7].................................11
2.7 illustration of five steps of detecting the flaws by using liquid penetrant testing [3] 13
2.8 Crack detection by using Magnetic particles testing [9]..............................14
2.9 Illustration of Acoustic Emission detection [11].....................................15
2.10 Electromagnetic spectrum [12].......................................................16
3.1 Optical sensor techniques.............................................................17
3.2 Illustration of multiple viewpoints technique [13]....................................18
3.3 Illustration of generating 3D image by using shape from shading [13]..................20
3.4 Illustration of active stereo with laser scanning technique [14]......................21
3.5 Illustration of using laser and camera for detect the surface [16]....................22
3.6 Illustration for some binary pattern [18].............................................23
3.7 Illustration of the camera and the projector coordinates [17].........................25
3.8 Illustration of 2D triangulation......................................................26
3.9 Illustration of 3D triangulation [18].................................................28
3.10 LiDAR scanning by using airplane [21]...............................................31
x


3.11 Illustration of phase shifting [23].............................................32
3.12 Illustration of calculating the Qi and Q2 [19]..................................32
4.1 The pipes that have been used for the experimental work............................33
4.2 The holes that have been introduced as defects.....................................34
4.3 The linear cracks in the pipes.....................................................34
4.4 Pipe with a small diameter (1.85 inches) and damage section........................35
4.5 Optical inner view of the E41 damage of GTI pipe sample..........................35
4.6 Two different cameras installed inside the pipe..................................36
4.7 The different defect types that have been captured by using the camera...........37
4.8 The laser source used in the sensor prototype....................................38
4.9 Simple camera and source bundle prototype........................................38
4.10 Some frames form the collected data: (a) with no defects and (b) with defects...39
4.11 Results after applying the thresholding to image frames shown in figure 4.10......40
4.12 Initial imaging result of 3D reconstruction using a stack of frames...............40
4.13 Fish-eye camera...................................................................41
4.14 the camera and the laser source installed in temporary prototype..................42
4.15 Complete ring images: a- ring with some deformation, b- Full ring.................42
4.16 The 3D reconstructed image with complete FOV......................................43
4.17 The prototype that have been used in this section.................................44
4.18 Illustration of one frame that has a defects......................................44
4.19 3D reconstruction................................................................45
4.20 structured light with using a projector to generate the pattern [24]............46
4.21 The lenses stacked inside the prototype.........................................47
4.22 The slide for the projector......................................................47
xi


4.23 The projection on the rough surface................................................48
4.24 Final shape of the projector.......................................................49
4.25 Projector and the camera connected to each other...................................49
4.26. a- the projection on the smooth surface, b-the prototype with blocked edges.......49
4.27. a- Illustration of one frame without defects, b- Illustration of a frame with squeezed
part......................................................................................50
4.28 Illumination of the white color has been reduced...................................50
4.29 3D reconstructed image with and without the defects.................................51
4.30 Some frames that show the squeezed parts inside the pipe............................51
5.1 Illustration of the circle which has fixed radius...................................53
5.2 Illustration of the ellipse [25].....................................................54
5.3 Illustration of 20 frames from collected data........................................55
5.4 Illustration of one frame with a defect and there are two lines to emphasize the ring
center....................................................................................56
5.5 Final result of the eccentricity has been calculated for 100 frames..................57
5.6 Illustration of the semi-major radius of all 100 frames..............................57
5.7 the region without defects but the shape is not circle..............................58
5.8 The eccentricity of multi frames without misalignment correction....................59
5.9 3D reconstructed image of number of frames without applying misalignment
correction................................................................................59
5.10 a- illustration of one frame that effected by misalignment and the shape is not a
circle, b- The same frame after applying misalignment correction and the shape is closer
to circle.................................................................................60
5.11 3D reconstructed image after applying misalignment correction.......................60
5.12 The eccentricity after applying misalignment correction.............................61
xii


1. Introduction
1.1 Background
There are a lot of developments in the last decades in infrastructures which made the
distribution of plastic gas pipes more complicated and widespread across the US. Plastic
pipes have some benefits that make this kind of pipes more sufficient than other types of
pipes. Corrosion and chemical resistance of plastic pipes are very high because the
conductivity of the plastic is minuscule and the electrolytic erosion has no impact.
Moreover, these pipes can be used to handle chemical solutions more than the other pipes,
so, plastic pipes can live for a long term. In addition to these benefits, the friction loss is
little because the inner surface of the plastic pipes is smooth then small power is required
to transmit the fluid. Flexibility also is one of the benefits when to compare it with metal
pipes installation.
However, some discontinuities occur to the plastic pipes which lead to prevent providing
reliable service. These discontinuities as a following; Manufacturing failures happen when
the pipe is not manufactured precisely and lead to having a crack or squeezed surfaces.
Construction problems: when the pipes not installed properly in the ground then the pipe
will be pressed. Tensile and external pressure: external pressure happens when the pipe is
buried in the ground and not symmetric load will be applied then the pipes will have some
bending. Tensile failure has four types which are [21] Ductile tensile, Brittle tensile,
Fatigue failure, and Compression failure
Because of these failures, diagnostic testing should be provided to inspect the failures and
give a sufficient discretion of the defects. There are two types of testing that can be used
to inspect these defects which are; Destructive Testing and Non-destructive Testing
1


Destructive testing mainly used to inspect the materials but will impact the characteristics
of these materials as will be described in chapter 2 then the tested materials will not be
useful. Therefore, Non-destructive testing have chosen to use in this thesis. NDE have a
different type of methods which will be clarified in chapter 2 as well. Visual and Optical
Testing is the method that used in this thesis. Laser scanning and structured light scanning
are the primary methods of new visual and optical testing. Laser scanning has been used to
inspect the deformation of the inner side of the pipes [2], In this thesis, pipes with 3 inches
will be tested by using laser scanning to generate a 3D image. A laser ring will concentrate
inside the pipe then a camera will capture the view and collect the frames to process them
by using a computer software. Moreover, structured light with multi colors multi rings will
scan the small pipe with 1.8 inches to generate 3D reconstructed image.
Figure 1.1 a- Illustration of ductile failure of PE pipe b-Typical slit type fractures [25]
1.2 Research Significance
This thesis is focusing on inspecting the inner surface of two different type of pipes. Laser
and structured light techniques have been used to investigate and analyze the deformation
of the collected data. Structured light biased on endoscope technique [4] will be used to
test 1.8 inches. The principle of using structured light is to send a pattern of light toward
2


the object then the illumination will be observed by using a camera. However, the main
contributions of this thesis can be as a following: Making some deformation in 3 inches
pipes to test them by using laser scanning technique. This step has three main categories
which are: Scanning by using small view angle camera, Scanning by using large view angle
(fisheye camera), and Generating 3D reconstructed image. In addition, multi rings multi
colors structured light will be explained to scan the pipes that have a diameter 1.8 inches.
And a new slide projector will be designed as well to generate multi colors multi rings
structured light. Then, 3D reconstructed image will be generated.
Finally, by calculating the eccentricity of each frame, the deformation can be explained in
other way to show the defects without using 3D visualization. The eccentricity also can be
used to reduce the misalignment that produced from the camera misalignment.
1.3 Thesis Outline
Chapter Two: the difference between destructive and nondestructive techniques are
discussed as well as the illustration of the different kinds of the NDE methods.
Chapter Three: this chapter has the principles of optical sensors techniques which are
divided into passive and active techniques. These methods will be discussed briefly
including to the laser and structured light scanning.
Chapter Four: All the experimental works that have been done in this thesis will be
presented as well as the 3D reconstruction of each generation.
Chapter Five: The deformation and the misalignment will be analyzed and discussed by
calculating the eccentricity.
3


Chapter Six: All the work will be concluded in this chapter in addition to the recommended
future work.
4


2. Nondestructive Evaluation (NDE)
2.1 Introduction
Destructive testing also is known as mechanical test [1], When some materials have tested
by destructive testing methods, the materials will be destroyed, or the characteristic of these
materials will be changed as illustrated in figure 2.1. Destructive testing can be Static or
dynamic, and this test is useful to evaluate: Yield point, Fatigue life, Hardness, Ductility,
Ultimate tensile strength, Elongation Characteristics, Impact Resistance, Toughness, and
Corrosion Resistance as illustrates in [1], Although destructive testing sometimes gives a
proper result, the materials that have been tested will not be useful anymore in same
purpose [1] [2],
Figure 2.1 sample that shows the destructive evaluation will affect the characteristic of
the spacemen [4]
NDE is techniques that have been used to evaluate structures or materials without
destroying or changing the system usefulness as destructive testing does [2] [3], Some
other definitions have been used for nondestructive evaluation which are: Nondestructive
Testing (NDT) or Nondestructive Inspection (NDI). In this chapter NDT, NDI, NDE will
be used interchangeably.
5


Destructive testing (evaluation) was the most known technique before NDE was invented.
NDE is not just one technique that has the equal modality to detect the deformations, but
different kind of techniques can be illustrated in various kind of obstacles [3], Visual and
optical testing, Radiographic Testing, Electromagnetic testing, Ultrasonic testing, Liquid
penetrant testing, Magnetic particle testing, Acoustic emission testing, and Infrared and
thermal testing are the NDE techniques that are used in different fields [2], NDE is not a
method that solves and fixes the cracks or deformations in the tested materials, but it helps
to detect these defects in order to give right solutions that contribute to enhancing the
efficiency of the materials. In the pipes companies especially the plastic pipes need to check
the cracks and the production quality to detect whether if useful to handle the servers or
need to have some maintenance. In this case, NDE methods should be provided to identify
and evaluate the materials without destroying them [4], Likewise, in medical field NDE
techniques are broadly used. For example: if a doctor needs to check the human gut, the
endoscope will be passed inside the patient body to see the problems. Then the treatment
will be prescribed according to the test result. Also, the X-ray can be used to detect the
sprain or the crack in the bones as the first step to give a proper explication. As a result,
NDE is not a simple method to use in one area or in specific problems, but it can be utilized
in any field because the idea of NDE is similar to what the human body use to sense the
things in their daily life. This type of the evaluation can be divided into several partitions:
NDE is used to study the materials properties as well as to their quality. The cost of the
production and maintenance can be reduced when NDE methods are used. The time of the
test can be reduced, and the size of the materials can be measured [2] [3], NDE will not
give a right evaluation when the wrong method is used to check the materials. As a result,
6


using a proper method of NDE will help to avoid or decrease the limitation that may happen
if the wrong method is applied.
2.2 NDE Methods
2.2.1 Visual and Optical Testing (VT)
Visual and optical testing is the most common method that has been used for a while [1]
[2], Also is known as the easiest way to examine the materials in addition to being the
initial step that implements before using any other NDE methods. The evaluation will lead
to getting wrong results if VT has been used in not a proper reason. For example, if the
material has some defects that are not showed on the surface, the VT will not detect these
defects because they located inside the material. Then by knowing the obstacle before
choosing the method is the first step to getting the right evaluation [3], VT is used to inspect
the materials that have some failing in such as corrosion, Holes, Cracks, and Blisters [1][3],
In this thesis VT is used in order to evaluate the cracks in PE and PVC pipes.
For old VT method, simple types of equipment are used to detect the flaws as illustrated in
figure 2.2; a just particular mirror can be sufficient to see the flaws. This simple method
has a lot of limitations that make it not useful when complicated objects have been tested.
For complicated objects, the human vision becomes harder to use [3], Therefore, the optical
and the computer vision should be used to evaluate this kind of failings. The new method
that is used the computer vision [3] to evaluate the materials is more beneficial. The laser
scanner and structure light are used to illuminate the object, and then a sensor is used to
capture the scene in order to generate 3D visualization and evaluate the cracks in the
materials. Laser scanning and structured light scanning will be discussed in details in
chapter 3.
7


Figure 2.2 Illustration of the old technique of using visual testing [5]
2.2.2 Radiographic Testing (RT)
RT is also one of the NDE methods that are widely used. After Wilhelm Conrad Roentgen
has discovered the X-ray in 1895 [2], the NDE started using this technique as a one of the
primary methods to check the discontinuities of the materials [2] [3], Figure 2.3 shows the
principle of the radiographic testing (x-ray). This method has been used in medical and
engineering experiments. The most important feature of this method is hidden flaws can
be detected and displayed in real time. As some other methods, RT has advantages which
are: The surface and the under surfaces can be scanned as well as hidden parts can be
detected, and most of the materials can be scanned by this method. RT also has some
limitations which have some health issue can be occurred because of the radiations that
come from the X-ray [3], Besides, a good experience is required to handle this method as
well as the expensive equipment that are needed [1] [3],
8


Ridialton sourc*
/ / i \ X
RKAtviAQ ModmiTi '
Figure 2.3 discontinuity detection by radiographic testing (x-ray) [5]
2.2.2 Electromagnetic Testing (ET)
Electromagnetic testing methods have been used to detect the cracks and flaw in the
nonferrous materials such as rod and wire. Figure 2.4 shows tested material surrounded by
the coil and alternated current is sent through this coil. Then the receiver coil detects the
current that have interacted with the material. Finally, the received signal is compared with
the reference signal to show the cracks [3], In ET imaging, ET methods are essential in
NDE area. Maxwell's equations are essential in all ET methods as well as covering a wide
range of electromagnetic spectrum from static and direct current. ET methods are
fundamental in order to detect the flaws in a case of dielectric material and conducting
materials as illustrates in [15],
Microwave imaging is one of ET applications in addition to eddy current and terahertz
imaging. In [15] Deng and Liu have divided the ET methods according to the wavelength
(from long wavelength to short wavelength). These methods based on this classification
9


are; static electromagnetic methods, Quasi-Static imaging methods, and high-frequency
time varying imaging methods. Each technique has been explained in [22],
Exc.utton Coil
Figure. 2.4 The principle of electromagnetic testing [6]
Figure 2.5 Illustration of using ferrous inclusion to detect the metal [6]
10


2.2.4 Ultrasonic Testing (UT)
Ultrasonic Testing is one of the methods that use high frequency, ultrasonic waves, and
mechanical waves to evaluate the materials. UT has been applied in both medical and
industrial inspection [2], Cracks and flaws can be recognized by using high frequencies
(0.5 to 15 MHz) [7], The primary process of using the UT is to send a signal using
transducer then if there is any crack or flaw in the object the signal will be reflected to the
receiver with different time [7] and magnitude as shown in figure 2.6. And then the signal
will be displayed and analyzed. UT just is needed one side to do the inspection (all surface
is not required to do the test). Also, UT has high accuracy as illustrated in [7], UT does not
have health issues same as RT does. UT uses portable equipment which makes this method
more useful. The surface and subsurface are sensitive to the test [3] [7],
UT is not beneficial for some materials according to sound transmission as well as it is not
useful for small, rough, and not homogenous materials. UT needs reference signal to
compare it with the received signal. Moreover, the surface should be accessible to transmit
the ultrasound signal. And the last disadvantage is UT needs advanced training comparing
with the other methods.
11


2.2.5 Liquid penetrant Testing (PT)
Liquid penetrant testing is one of oldest and simplest method in the NDE technique because
this method does not use any new technology to detect the flaws and cracks. The instructor
just needs simple tools to evaluate the sample. But non-porous feature should be existent
to ensure the test is sufficient [8], There are several steps to implement PT test which is
started from cleaning the surface from any kind of oils or dust. Then the penetrant material
is applied to the sample. There are two types of the penetrant material which are: visible
penetrant and fluorescent penetrant. After implementing the penetrant to the material, the
dwell time should be considered before removing the penetrant. Next step is to eliminate
the penetrant. The penetrant can be moved by water washable, post-emulsifiable,
lipophilic, solvent removable, or post-emulsifiable, hydrophilic [3] [8],
The last step is to detect the cracks and flaws by using the developer as shown in figure
2.7. This action takes some time to detect the cracks spatially for the materials that have
small cracks. PT it is portable technique and the cost is low as well. The order advantages
are PT can detect the fine cracks and different kind of materials can be identified by this
method.
This method as mentioned is one of the easiest methods to identify the flaws in some
material. But comparing with the other NDE methods, PT still has limitations that make
this method not sufficient in some cases. Such as just nonporous materials can be inspected,
pre-cleaning can hide some of the defects as illustrated in [8], flaws should be on the
surface, inner flaws cannot be detected, and removing the penetrant after doing the test is
required.
12


I Pie-cleaning,
Remove did
and dust from tin?
surface with Remover.
2. Penetrant
Application.
Apply Dye Pene-
trant and leave
it as is for five
to ten minutes.
-^iSSS

Penetrant
Removal
Remove excess
surface- Dye Penetrant
-with Remover.
Developing
5. Inspection
Apply Developer
uniformly over
the surface.
Defects will be
found in a bright
red indication.
Figure 2.7 illustration of five steps of detecting the flaws by using liquid penetrant
testing [3]
13


2.2.6 Magnetic Particle Testing (MT)
The basic concept of using MT method is taking advantage of the magnetic field of the
materials. MT is adopted to investigate a diversity of product forms containing forgings,
castings, and weldments. Before implementing MT, the characteristics of the material
should be identified as illustrated in [2], Just ferromagnetic materials can be inspected by
using this method. Moreover, the flaw should be on the surface of the tested material to be
able to recognize. Otherwise, this method will not be useful.
When the material is magnetized, the flux lines will travel from south to the north pole.
Since the material has a crack, the flux will exit the material and then reenter (flux leakage)
again to the material as shown in figure 2.8 [9], An iron particle is used to check the
leakage. The Iron will be attracted by the flux leakage, and then the crack can be defined
easily by knowing the attracted area.
MAGNETIC FIELD LINES MAGNETIC PARTICLES
- 4
14


2.2.7 Acoustic Emission Testing (AE)
The fundamental principle of the acoustic emission is to use elastic waves. The acoustic
emission source is used to sent the elastic waves toward the material (that is needed to test)
via a transmission media. The wavies change to the electrical signals in order to be
magnified and processed. Then the defects can be displayed on the screen as illustrated in
[10].
Figure 2.9 Illustration of Acoustic Emission detection [11]
2.2.8 Infrared and Thermal Testing (IR)
Infrared and thermal testing is also known as infrared vision [2], Infrared waves located in
the electromagnetic spectrum as shown in the figure 2.10.
For infrared waves, there are Mid Wave IR (MWIR) and Low Wave IR (LWIR) which
both of them have a different wavelength. The wavelength of LMIR is between 7 and 14
pm, but the wavelength of MWIR is between 3 and 5 pm. By using the infrared sensor to
detect the surface. When thermal waves are applied on the material, the cracks and flows
will interact with the waves differently comparing with areas that do not have defects. By
measuring the difference of the thermal waves, the defects can be determined. Moreover,
15


IR uses expensive equipment such as a thermal camera which makes this technique has a
cost limitation comparing with efficient and cheaper methods [2],
III

80%
atmospheric
transmission
NIR SWIR MWIR
Low
atmospheric
transmittance
window
l
0.01 0.4 0.7 1 2 3
//
cr/
V
Non-thermal
/
/
Figure 2.10 Electromagnetic spectrum [12]
1000 pm
16


3 Optical Sensors
3.1 Introduction
Optical sensing is one of the NDE techniques that have been improved significantly in past
few years. There are several ways under optical sensing that produce a 3D image. However,
two different techniques will be the main categories which are passive acquisition and
active acquisition, in which laser and structured light will be classified under active
acquisition. The following chart is shown in figure 3.1 illustrates the main categories of the
Optical sensor.
Optical sensor
Passive Acquisition
Active Acquisition
Multiple Single View Active- Triangulation Time-of-Flight
viewpoints point stereo
Figure 3.1 Optical sensor techniques
3.2 Passive Acquisition
Passive Acquisition is one of the methods that can be used to get 3D optical images. This
technique is called passive because the data is collected by using the camera without using
an active source, such as lasers to provide strip line or by using structured lights. Multiple
viewpoints and single viewpoint techniques are classified to be the primary methods for
the passive acquisition.
17


3.2.1 Multiple Viewpoints
This technique is used to take the image from different viewpoints/angles. Two cameras
or more can be used to collect the data from various angles. By using multiple cameras, the
technique is called stereo vision. There are different kinds of stereo with a different number
of the cameras that have been used. For example: if two cameras have been used to capture
the image the technique is called binocular stereo. Similarly, trinocular stereo is when three
cameras have been used. Also, if a single camera has been used to take the image at the
same point but from different locations and time, this technique is called structure from
motion. By processing the images that have been taken from the different locations or
different cameras, the 3D rays will be determined. Finally, from the 3D rays, the 3D
position of the point in the scene can be determined. One illustration of the multiple
viewpoints is shown in figure 3.2.
Figure 3.2 Illustration of multiple viewpoints technique [13]
18


3.2.2 Single Viewpoints
In this technique, the captured image does not come from multiple cameras or the camera
motion. The captured image, in this case, is taken by using the object details, for example,
the texture of the object, shading, or focus, etc. Figure 3.3 shows that shape from shading
has generated the 3D image. This technique depends on the reflection from the object. The
pixels in the reconstructed image illustrate the intensity of the reflection that comes from
the object shade, and by using regularized surface fitting, the 3D image can be
reconstructed. Overall, the multiple viewpoints method is more accurate and more efficient
than single viewpoint method. This techniques limitations and the enhancement of the
reconstructed image are described in [1], Shape from shading technique is known as
Photometric stereo. The Main steps to generate a 3D image by using this technique is that
taking more than one image at the same point but the illuminations for the scenes are
different. The other technique is called shape from focus. Taking two images from different
depths of field is the main idea of processing 3D image, and the approach is described in
[3].
In summary, single view technique is not as good as multiple viewpoints technique in terms
of speed and regulation. Therefore, multiple viewpoints technique is more commonly used.
However, the passive acquisition is not suitable for this project. Active acquisition will be
the tool for making the 3D imaging which will be reviewed and discussed in the following
section.
19


<1
Figure 3.3 Illustration of generating 3D image by using shape from shading [13]
3.3 Active Acquisition
Active Acquisition has different techniques to capture the viewpoints comparing with
passive techniques. As illustrated in the passive techniques, captured image does not need
structured light or laser strip to collect the data. However, in Active Acquisition, optical
detectors such as camera need to be used to detect the spot. Besides, the active sources such
as laser source or structure light should be considered to complete the imaging process.
3.3.1 Active-Stereo
Active stereo has a similar idea with multiple viewpoints. In this technique, a light source
with unique features has been used to replace the function one of the cameras in the
multiple viewpoints method. Figure 3.4 illustrates the principle of the Active-stereo method
with laser scanning. The light source focuses on the scene, and then the camera captures
the view. There are different kinds of the light sources. Structured light and the laser are
two of the most known sources. These techniques will be discussed in the following
sections. Active stereo is more common than passive techniques because of the high
20


resolution and the data that have been taken by the laser, or structured light are more
accurate. However, to generate a 3D image by using a camera and light source, the
triangulation should be considered to calculate the depth of the scene, which makes the
method mathematically rigorous and challenging.
Laser source
*
Figure 3.4 Illustration of active stereo with laser scanning technique [14]
3.3.1.1 Laser Scanning
A lot of applications have used laser scanning to get a 3D image. Laser scanning uses a
strip of laser light, or it can be a circle-shaped source to scan the inner surface of an object,
e.g. pipelines in this project, which will be discussed in details later. Basically, a camera
with high resolution uses as a sensor to capture the scene. Both the camera and the laser
source functioning as one unit to produce the 3D image. Figure 3.5 illustrates how laser
and camera work together.
The distance between, the camera and the laser source, and the illumination of the object
have a particular relation called triangulation to create 3D images. When the laser
21


projects the light on the object, the laser light will capture the object shape. Then by
knowing the camera and laser location, the depth of the object can be easily defined.
Figure 3.5 Illustration of using laser and camera for detect the surface [16]
3.3.1.2 Structured Light
As discussed in laser scanning section, structured light is similar to laser technique in setup;
however, the structured light projector has been used in this technique instead of the laser
source. The structured light technique has been utilized for a lot of applications, for
example, industrial and medical testing. The principle of structured light is shown in Figure
3.6, which is depending on the relative positions of camera and projector. In our prototype,
the projector is connected to the computer and will be controlled to generate pre-designed
patterns. There are a different kind of patterns that have been used for the structured light
such as: Spatial patterns, temporal patterns, Color patterns.
22


Figure 3.6 Illustration for some binary pattern [18]
When the projector projecting the light on the scene, the camera will capture the
illumination from the object. Likewise, the light will take the object shape. By knowing
the intersect point between projector light and the camera view, the 3D image can be
calculated.
The 3D coordinates for the point as shown in the figure 3.7 is given by Xc = [Xc YCZC 1 ]T
Then the image that have seen by the camera can be determined by the following:
*c = PCXC...................(Eq3.1)
And the Pcis the matrix that present the camera perspective
~vx k ^cO 0
Pc = 0 Vy yc o 0
-0 0 l 0
23


Also for the projector, the 3D point in the object according to the projector coordinates
Xp [Xp Yp Zp lj
Then back projection is going to be xp = Pp Xp
Vp 0 xpo 0
0 0 1 0-
By capturing the image, the relation between the image that have been captured by the
camera and the projector view has some factors such as; rotation and translation. Then the
relation can be written as
Xp = MXC....................(Eq3.2)
M = R()RaRpT.....................(Eq3.3)
When Rq, Ra, Rp, T are the matrices for the rotation in different axis and the translation
Then
xp = PpMXc...................(Eq3.4)
By multiply PpM new matrix will be defined when rqand r2 4D row vectors
xpr2Xc = r{Xc..................(Eq3.5)
xpr2Xc rxXc = 0.................(Eq3.6)
(xpr2 ri)Xc = 0.................(Eq3.7)
From equation (Eq3.1) and (Eq3.7)
24


Then
Pc
-V* -y V
LJcprZ '1
Xc
(Eq3.8)
If
Pc
-V* y V
xpI 2 '1
= H
(Eq3.9)
II Xc = xc .................(Eq3.10)
Finally the 3D position for any point in the object can be calculated by knowing H matrix
Xc = H~1xc+..................(Eq3.11)
Figure 3.7 Illustration of the camera and the projector coordinates
25


3.3.2 Triangulation
Triangulation is an essential step to building a 3D image because the relation between the
camera and the light source should be defined to calculate the depth. Triangulation is used
in the passive and active acquisition. However, in a following steps triangulation process
of the active acquisition will be described. Figure 3.8 shows a 2D triangulation system [23],
And the approach is described in [24], The distance between the camera and the light
source referred as b. The angle of the light source is a. and the angle between the camera
and the baseline is p. P is the intersect point with the camera view and the light source
projection. L O P are the points that determine the location of P in the object.
Distance d can be calculated by knowing the angles a (3 y
By using sine law
26


b d sin(a) sin(y) Then (Eq3.12)
, (l)*sin(a)) a = sin(y) (Eq3.13)
When sin(y) = sin(n + /?) (Eq3.14)
. l)*sin(a) a sin(a+^) (Eq3.15)
Now the coordinates of the point P can be determined by the /? which is controlled by the
projection of the light source on the project.
P = (d. cos /?, d. sin (3) (Eq3.16)
When x = d. cos (3 (Eq3.17)
h = d.sin/? (Eq3.18)
Similarly, to find 3D image points that described in [24] and illustrated in figure 3.9, the
3D coordinators can be determined by knowing a and b.
X Z Y
From ray theorem = = -...................(Eq3.19)
/ = -./' X Z can be defined by a angle. ... (Eq3.19)
Z = tan a. (b X) (Eq3.20)
-. / = tan a. (b X) (Eq3.21)
27


(Eq3.22)
x
~.f = b. tan a X. tan a
X
X
-.f + X. tan a = b. tan a
X
f
- + tan a
X
.tana
X =
b.x.tan a
f+x.tan a
.....(Eq3.23)
.....(Eq3.24)
(Eq3.25)
Then the coordinates will be as following:
X =
b.x.tan a
f+x.tan a
b.y.tan a
/+x.tana
z =
b.z.tan a
f+x.tan a
(Eq3.26)
Figure 3.9 Illustration of 3D triangulation
28


3.3.3 Time-of-Flight
Laser and structured light use visible light that can be seen by human vision but there are
other signals e.g. (invisible light) can also be useful for scanning. In the applications that
use invisible light, the Time of Flight (TOF) is one of the representations that used to define
the depth or the distance. TOF has been illustrated in [19] [20] and the signal that will send
from the source is modulated, then this signal will be reflected toward the receiver. There
are three main divisions of TOF. Each one has a different approach with the transmitted
signal [20],
3.3.3.1 Direct Time-of -Flight
This technique uses discrete pulses to send a signal and receive it by a sensor. According
to the time difference between the transmitted and reflected signal, the distance can be
calculated easily by using the following formula.
d=Y.....................(Eq3.27)
When c is the light speed
c= 299,792,458 m/s
Direct time of flight is difficult to implement as illustrates in [20], when the accuracy is
needed to be in centimeter level. High clock speed is required in order to increase the
accuracy [20],
3.3.3.1.1 LiDAR
Light Detection and Ringing (LiDAR) has been developed since the 1960s [21], LiDAR is
one of the techniques that depends on TOF process. A beam of a laser is adopted as a source
for LiDAR. This laser beam uses to measure the distance between the obj ect and the source.
29


However, LiDAR is known as a fast scanner because this technique can collect the data
more quickly than the other techniques. There are different kinds of the LiDAR some of
them use fixed LiDAR to collect the data, and the others use moving LiDAR. Infrastructure
engineering is one of the applications that dependence on the fixed LiDAR as illustrates in
[20], Airborne LiDAR, Mobile Terrestrial LiDAR, and Static Terrestrial LiDAR are the
main types of LiDAR technique [21],
Airborne LiDAR is used for the applications that need to scan from particular high, and the
others are used on the ground. As discussed in the TOF, the speed of light and the reflection
time are required to calculate the distance. GPS is needed in order to define the X, Y
positions [22], Therefore, the accuracy of the LiDAR system depends on the GPS accuracy,
not just the laser reflection. Figure3.10 shows an image has taken by LiDAR scanner. This
system has some significant equipment functioning together to generate the 3D image such
as Laser scanner, GPS, Data storage and management systems, High-precision clock, GPS
ground station, and IMU Inertial navigation measurement unit [21],
The laser frequency that is applied in LiDAR devices is around 50 kHz to 200 kHz [21],
Also, the wavelength of the laser beam is different from the system to another. In Doppler
LiDAR as discussed in [21] the wavelength between 1500 and 2000 nm (infrared). In
contrast, terrestrial mapping the wavelength between 1040 and 1060 nm (near infrared).
3.33.2 Range-Gated Imaging
Range-gated Imaging is not same as direct time-of-flight in calculating the distance. Range-
gated Imaging as illustrates in [20] depends on signal integration in order to calculate the
distance. A CCD or CMOS are used to capture the energy then the distance can be
measured according to the different intervals of the energy ratio. The dynamic range of the
30


CCD and CMOS is limited [20], Therefore, the strong light in the surrounded area will
affect the measurements.
Figure 3.10 LiDAR scanning by using airplane [21]
3.3.3.3 Phase Difference Measurement
The phase difference between transmitted and received signal can be used to calculate the
distance as discussed in [20], The transmitter sends a pulse as illustrates in figure 3.11 then
this signal will be reflected back to the receiver. The reflected signal has a phase difference
comparing with the original. The receiver has a feature that works to change the received
pulses to electrical current as described in [19],
The time period between the source and the object is At as same as the time between the
reflected signal and the receiver. Then the depth can be calculated by using following
formula.
d
2
1
C
(Eq3.28)
31


Signal seri out T77. rs** //j Si jial a ait out 1
Sijjial revived Corr elation / J y//& {A
w v//, Si trial received _jn
Correlation n n r
-ii ii i-
Figure 3.11 Illustration of phase shifting [23]
When c is speed of the light
Qi and Q2 are the electrical charge for pulse signal determined in same time as the
fallowing figure
Integration Time
<---------------------------------------------------------v
Qi
Q,
L
I_____________________________I
Figure 3.12 Illustration of calculating the Qi and Q2 [19]
Light Source
Reflection
Cl
C2
32


4. 3D Pipe Inner Wall Reconstruction
4.1. General Overview
As known NDE technique will be used in this research and specifically the Visual and
optical testing method will be considered a main method. Also, the structured light is
considered as our light source in this step. Therefore, the designing of the new prototype
that uses structured light with multi-rings multi-colors will be explained in this section
4.2 Materials and Preparation
4.2.1 The Defects and Cracks in the Pipes
In the beginning, 3 inches diameter pipes are considered as the specimen. Figure 4.1 shows
the pipe that has been used in the experimental work.
Figure 4.1 The pipes that have been used for the experimental work
Then some defects have been made in the pipe as illustrated in figure 4.2. Through hole
defects with different diameters were introduced. In the white pipe specimen, the smallest
hole in the pipe has the diameter of 0.0787 inches and the biggest hole is 0.314 inches.
Also, there are additional holes with 0.157 and 0.197 inches diameters. Moreover, some
33


screws have been installed to mimic different defect types. Figure 4.3 shows the cracks that
have different directions in the pipe and the width of the crack is 0.039 inches.
Figure 4.3 The linear cracks in the pipes
34


Figure 4.4 Pipe with a small diameter (1.85 inches) and damage section
The inner view of the damage area is shown using commercially available optical camera
in Figure 4.5.
Figure 4.5 Optical inner view of the E41 damage of GTI pipe sample
4.2.2 Scanning without Light Source
Before doing the laser scanning or structured light scanning, passive scanning is considered
to show the defects without any light source. However, two different cameras that have
been used to scan the inner side of the pipes as shown in figure 4.6. These cameras have
35


different diameters. The diameter of the smallest camera is 0.23 inches and the diameter of
the largest camera is 0.62 inches. The camera is installed in the special prototype that has
been produced by using 3D printer in our lab. After installing the camera and moving it
inside the pipe, a video will be recorded to capture all inner surfaces in the pipe.
Figure (4.7-a) illustrates the cracks that have been captured by using the camera. Also, in
the figure (4.7-b) shows the holes that have been introduced in the pipe.
Figure 4.6 Two different cameras installed inside the pipe
36


(a) (b)
Figure 4.7 The different defect types that have been captured by using the camera
4.3. Single Ring Scanning
In this part of the research, the single ring is considered to be the light source of the scanner.
Laser source with specific red ring has been used to detect the cracks and the deformation
as shown in figure 4.8. The relation between the diameter of the ring and the distance
between the ring and the laser source is (1:1). This means that if the distance between the
laser and the scene is 1 meter then the diameter of the ring on the scene will be 1 meter.
37


4.3.1 Scanning with Small View Angle Camera
The camera that has been showed in the figure 4.6 is used with the laser source to scan the
pipe. Figure 4.9 illustrates the prototype that has been used for scanning. Actually, the
camera in this prototype will be connected to the laser source and shifted few centimeters
back. This shifting because the view angle of the camera is small. So some shifting is
needed to make the camera capture the rest of the laser ring. By moving the prototype with
constant speed, the camera will collect the data and will be saved in the computer.
Figure 4.9 Simple camera and source bundle prototype
38


As shown in figure 4.10, the laser ring is not completed because the camera will capture a
part of the laser source. As a result, some of the scene will be blocked. This kind of missing
is called shadowing. This shadow area blocks about 25% of the whole scene. So this kind
of missing is one of the limitations that should be considered to solve in the next sections.
However, the laser ring will take the inner surface shape of the pipe. Figure (4.10-a)
illustrates that the laser ring does not has any deformation. On the other hand, figure (4.10-
b) shows some deformation that happened to the laser beam. The changes in the laser light
shape will be considered as indications of defects that we want to see in the reconstructed
images.
(a) (b)
Figure 4.10 Some frames form the collected data: (a) with no defects and (b) with defects
Then the collected data were processed by recon algorithms. Some of image processing
tasks will be considered to generate the final 3D image. In the beginning, thresholding and
some digital filtration will be applied, then the result will be shown in figure. 4.11.
Moreover, digital image enhancement techniques have been used to enhance and detect the
deformation such as image closing, dilation, and other morphological process. Finally, after
repeating this process for all the frames, the 3D image reconstructed can be easily
39


generated. Figure. 4.12 shows the 3D reconstructed image. There are some holes in the
reconstructed image that show the defects that already have been done in the first section.
Clearly some of the pipe will not be reconstructed because of the shadows that have been
done by the laser source. As result, the next section will discuss how we can get full 3D
reconstructed image.
Figure 4.11 Results after applying the thresholding to image frames shown in figure 4.10
Figure 4.12 Initial imaging result of 3D reconstruction using a stack of frames
40


4.3.2 Scanning with Large View Angle Camera
In this section we will repeat the same 3D reconstruction but with different prototype.
Actually, the missing part in the previous 3D image will cause some problems. For
example, if there are any defects or cracks in the missing area, the 3D reconstructed image
will not show this defect. So, some enhancement should be done to avoid this kind of
problems.
Using large view angle camera is one of the solutions that make the reconstructed image
better. The difference between this camera and the previous one is the view angle. In the
old one the view angle is about 45 degree, but in the new one is 180 degree that makes the
whole scene can be detected by putting the camera and the laser source in the same base
line. The new camera is called fisheye camera. Convex lenses have been installed to this
camera to make the view angle bigger. Figure 4.13 shows the fisheye camera that has been
used in this experimental work.
Figure 4.13 Fish-eye camera
In the last prototype the camera was contacted to the laser source but in the new prototype
the camera and the laser source will be disconnected and they will be in same base line.
Because the camera has 180 degree view angles, all the ring will be captured without
41


missing any part. The new prototype can be illustrated in the figure 4.14. Then, the
prototype will be passed in to the specimen to collect the data. However, the scanning will
be by human hand. So, some misalignment will be considered. After that, the same 3D
reconstruction process will be repeated. Then by taking a few frames from the recorded
video, the laser light inside the pipe will be as shown in figure (4.15 a-b). In the figure
(4.15-a) the ring is completed and the laser light does not have any deformation. This
indicates that the areas that have been covered by this light are still without any defects.
On the other hand, figure (4.15-b) shows the small deformation in the laser light.
Figure 4.14 the camera and the laser source installed in temporary prototype
(a) (b)
Figure 4.15 Complete ring images: a- ring with some deformation, b- Full ring.
42


According to this deformation, this frame will represent the area that have the defect in the
pipe. Likewise, by stacking all the frames and generate the 3D image, the defects will be
clearly appeared in the final image figure 4.16.
100 200 300 400 500 600 700 800
Figure 4.16 The 3D reconstructed image with complete FOV
Actually, the laser and camera are taking a large space inside the pipe that make the
minimization of the prototype more complicated. Different prototypes are being developed
to minimize the size of the sensor and will be discussed in next section.
43


4.3.3 Scanner and the Sensor in Opposite Direction
As mentioned in the previous section that the laser source and the camera will take more
space if they have been installed in the same baseline. Therefore, in this section different
prototype will be illustrated. Although this prototype will reduce the size of the scanner,
the image will not be full ring. Some missing part will be shown in the collected data
because of the metal that has been used to connect the laser source and the camera. A part
of the light will be blocked and will not be recorded by connecting the prototype as shown
in figure 4.17. Therefore, when moving the prototype inside the pipe, the scene will be as
shown in figure 4.18.
Figure 4.17 The prototype that have been used in this section
44


Clearly, the defects can be illustrated as missing points in the laser beam. As a result. After
processing all the frames, the final 3D reconstructed image will be as shown in figure 4.19
Figure 4.19 3D reconstruction
4.4 Multi-Rings Multi-Colors Scanning
As has been discussed in chapter 3, structured light is one of the methods that has been
used to generate a 3D image by using the active technique. To produce a pattern that is
needed to illuminate the scene, a projector is required to do this step. In typical 3D imaging
by using structured light, the projector will be similar to one that is shown in figure 4.20.
However, this kind of projectors will not be sufficient inside the small pipes. As a result,
45


another device is needed to do this work in order to fit inside the pipe. A small slide
projector has been developed to provide the particular pattern.
Figure 4.20 structured light with using a projector to generate the pattern [24]
4.4.1 Projector Prototype
The main idea of using the slide projector is to emit a pattern with different colors on the
object to obtain different illumination areas. As a result, a small slide projector has been
designed to generate the colors instead of using the regular projector. Moreover, to create
this projector, some lenses are needed. The lenses that have been chosen for this step have
a diameter (1.5 inches). Therefore, the size of the projector will go for the small pipes that
we need to scan. Then, these lenses will be stacked with a certain way inside a prototype
that has been designed by using the 3D printer. One achromatic lens, two double convex
lenses, and two convex lenses are used as shown in figure 4.21. Every lens has a focal
length that should be considered when the designer stacks the lenses together. The main
idea of using the Achromatic lens is to reduce effects of chromatic and spherical aberration.
The other lenses are used to magnify the image and increase the resolution. The slide that
is used in this prototype has multiple rings as illustrates in the figure 4.22. Moreover, the
46


power of the light source is strong enough to generate the light that is needed to emit the
pattern.
Figure 4.21 The lenses stacked inside the prototype
Figure 4.22 The slide for the projector
However, the slide has a two different colors (red and green) as shown in figure 4.22. There
is a black area between the two rings just to separate the edges. This separation can be
removed in the future after increasing the resolution. Figure 4.23 shows the pattern on the
rough surface to clarify the edges.
47


The length of the projector is 7.87 inches, and the largest diameter is 1.77 inches which
can be used to scan the pipe that has 1.85 inches. Also, the distance between the light source
and the rings when are projected inside the pipe is 1.02 inches. Then the next step is to
connect the prototype and the camera as shown in figure 4.25. The camera that has been
used in this step is the one that has small view angle. Because the pipe has a small diameter,
small view angle camera can be used instead to the large view angle camera. As shown in
the figure the distance between the light source and the camera is 5.51 inches.
Figure 4.23 The projection on the rough surface
By moving the prototype inside the pipe, the data will be collected by using the camera.
Each frame has some features that came from the structure of the pipe. Figure 4.27 (a)
illustrates some frames that have been taken from the data. Clearly, the shape of the rings
has been changed as shown in figure 4.27 (b). This changing shows the squeezed parts in
the pipe. As a result, the reconstructed image will have the same shape of the corrupted
image. But in this case, there is some illumination that has been generated from the extra
light that comes from the lenses. To block the extra light, a black tape has been used to
cover the edge of the lens as shown in figure 4.26 (b).
48


The difference can be easily illustrated in figure 4.28, and the colors clearer after blocking
the white light. Then, the algorithm will process each frame in order to generate the 3D
reconstructed image as shown figure 4.29.
Figure 4.24 Final shape of the projector
Figure 4.25 Projector and the camera connected to each other
49
iV- v


Figure 4.28 Illumination of the white color has been reduced
A color segmentation is done as a first step in the algorithm to generate the 3D image. The
red color is segmented and reconstructed as shown in figure 4.29. Also. Can be clearly seen
the one of the reconstructed images is not regular ring shape but it is twisted. The twisted
area in the image illustrates the squeezed part inside the pipe. Moreover, figure 4.30 shows
just a few number of frames that show the pressed area.
50


Figure 4.29 3D reconstructed image with and without the defects.
Figure 4.30 Some frames that show the squeezed parts inside the pipe
51


5. Discussion and Analysis
5.1 Deformation Detection
3 D visualization is an efficient method to emphasize the deformation in the pipes. But in
order to clarify the deformation by checking the frames if in a circle shape or not, the
eccentricity can be calculated to test the circularity. In this chapter, the eccentricity will be
calculated, and the misalignment correction also will be provided in order to minimize the
camera misalignment.
5.1.1 Eccentricity Calculation
As known the circle has a fixed diameter or radius in all directions as shown in figure 5.1.
This radius is measured from the center of the circle to the edge. Therefore, the distance
(a) equals to distance (b). On the other hand, in the case of the ellipse, the (a) and (b) are
not equal.
The ellipse has two different radiuses which are called semi-major and semi-minor. Semi-
major denoted as (a) and indicates to the larger radius, and the other radius (semi-minor)
denoted as (b). As seen in figure 5.2 there are two focal points which FI and F2, and the
distance between the focal point and the center is called linear eccentricity. Then by
choosing any point on the ellipse (p) and calculate the distance between it and the focal
points we can get the semi-major axes as following:
PF1 + PF2 = 2a...........................(Eq 5.1)
And
f2 = a2 -b2.............................(Eq 5.2)
Then by knowing (a), (b) and (f), other term can be calculated which is called eccentricity
and denoted by (e). In case of ellipse, (e) will be between 0 and 1 (0 < e > 1).
52


(Eq 5.3)
e
r
a
e =
Va2-i)2
a
e =
a2-b2
a2
e = 11
(Eq 5.4)
(Eq 5.5)
... (Eq 5.6)
For example if we have a=2.5 and b=2, then
' = J1 ~ (-hf......................... e = 0.6
Which is less than 1
Also if a=b=2.5
e=J1 S)2............................. e = 0
Which is in case of circle
As a result, by calculating the eccentricity for each frame, we can define if the frame in
shape of circle or not.
Figure 5.1 Illustration of the circle which has fixed radius
53


Figure 5.2 Illustration of the ellipse [25]
5.1.1.1 Eccentricity of the Area with Free Crack Deformation
In this section, the eccentricity formula will be applied for some frames that do not have a
deformation to emphasize whether if it has a circle shape or not. Because of some other
effects, the eccentricity may not be 0 in the case of free deformation area. This issue is
called misalignment and will be discussed in separate section. However, lets consider the
misalignment is negligible, and just focusing on the large variation of the sequence. Then
the defects can be easily illustrated. Figure 5.3 illustrates the eccentricity of 20 frames
without defects and the variation is not significant. The eccentricity in figure 5.3 varies
between 0.26 and 0.32.
By taking a special case to check the eccentricity. And frame number 60 has been chosen
then the semi-major radius is 200 (number of pixels). And semi-minor radius is 193 pixels
then
54


e =
(Eq 5.9)
e = 1
/193V
-(200) =-
2622
It is still not zero but this result will be compared with the results in the following section
5.1.1.2 Eccentricity of the Area with Squeezed Parts
For the frames that have defects, the eccentricity will be much bigger than the other that
have illustrated in the previous section. Therefore, by picking some frames that have
defects, and lets say ten frames (120-130). One of the frames is shown in figure 5.4.
Vertical and horizontal lines are added to the image to illustrate the center of the ring.
However, the eccentricity will be calculated for frame number 125. Then by calculating
semi-major radius which is 217 and the semi-minor radius will be 185
e = 1
-(D2
55


The eccentricity has been increased comparing with the previous result. To emphasize the
difference between the free crack area and squeezed part the eccentricity will be calculated
for all the frames as shown in figure 5.5. In the beginning, the eccentricity will be
fluctuating around small value then suddenly will start increasing until reaching the pick
value (0.58) after that will decrease until start fluctuating around small value again. From
this values, the deformation can be easily defined by the eccentricity.
Figure 5.4 Illustration of one frame with a defect and there are two lines to emphasize
the ring center
56


0.6
200 250 300 350
The distance(mm)
Figure 5.5 Final result of the eccentricity has been calculated for 71 frames (17.6 cm)
Figure 5.6 Illustration of the semi-major radius of all 71 frames (17.6 cm)
57


5.2 Misalignment Correction
The misalignment is one of the problems that are faced when the data has been analyzed.
This issue makes the reconstructed image not clear enough to emphasize the defects,
therefore, in this section the misalignment correction will be applied to one of the data that
does not have defects but the eccentricity is large because of the misalignment.
Figure 5.7 the region without defects but the shape is not circle
Moreover, if the eccentricity algorithm was applied to a couple of frames, the output will
not be zero as mentioned, the eccentricity will be around (0.6359) as shown in figure 5.8.
Also, the 3D reconstructed image will be as figure 5.9. As a result, the correction should
be applied to decrease the eccentricity to be closer to zero. This correction depends on the
semi-major radius, semi-minor radius, and the orientation of the ring. These parameters
58


used to determine the (tform) matrix by using affine transformation. Then the image will
be reshaped and squeezed to circle shape as shown in figure 5.10-b.
1 r j r i r T 1 1 1 T 1
0.9 -
0.8 -
0.7 -

O 0.6 -
k_
c
8 0.5 -
O
0.4 -
1-
0.3
0.2 -
0.1 0 -i 1 1 1
10 20 30 40 50 60 70 80 90 100 110 120
The distance(mm)
Figure 5.8 The eccentricity of multi frames without misalignment correction
o o
Figure 5.9 3D reconstructed image of number of frames without applying misalignment
correction
59


Figure 5.10 a- illustration of one frame that effected by misalignment and the shape is
not a circle, b- The same frame after applying misalignment correction and the shape is
closer to circle
Figure 5.11 3D reconstructed image after applying misalignment correction
60


The eccentricity
Finally, the eccentricity of the same frames after applying misalignment correction will be
around (zero) which is what we expected to have and figure 5.12 illustrates the eccentricity
after misalignment correction.
Figure 5.12 The eccentricity after applying misalignment correction
61


6. Conclusion and Future Work
6.1 Conclusion
Optical sensor and NDE techniques are effective methods to use for studying the material
discontinuities especially pipelines inner defects. Visual and optical testing are powerful
to detect the deformation inside the pipes, but by using active acquisition, the result will be
more beneficial. In chapter 2 the NDE methods have been discussed, and VT has been
chosen as the primary method of this thesis. VT method depends on computer vision to
detect the deformation, but the accuracy of this approach depends on the collecting data.
Therefore, the optical sensor has been discussed in chapter 3. The optical sensor has two
main techniques which are, passive acquisition and active acquisition. Both of methods has
been explained briefly, and active acquisition has been selected as the main method of this
thesis. Laser and structured light are classified under active acquisition as the methods that
have chosen to do the experimental work.
However, two different kinds of pipes have been tested to emphasize a different kind of
issues. First results were about a detection of different size of holes inside 3 inches pipeline,
and the results can be organized as a following:
Scanning with small view angle camera prevents to detect all the inner surface, and some
of the important data are not captured. Scanning with large view angle camera (fisheye
camera) captures all the scene, and the defects can be easily detected. One of the limitations
in this type of scanning is the color dependency, which means the accuracy of the detection
depends on the color of the laser and the pipe. The other limitation is the misalignment
when the prototype has some unexpected tilting because of using human hands while doing
62


the scan. In chapter 5, the misalignment problem has been reduced when the eccentricity
and image transformation have been done to the collected data.
The structured light technique has been used as a second method to emphasize the
discontinuities. In this part, a new structured light projector has been developed to generate
a multi-colors multi-rings structured light. Some lenses have been used in a certain way
with a pre-designed slide to build the projector. Then, a new kind of pipe has been presented
to test this method; all the steps has been explained in chapter 3 as well as the results of 3D
reconstructed image. The defects that have been examined is squeezed part from PE pipes.
The benefit and the limitation of this method can be introduced as a following:
Two different colors have been used to detect the squeezed part, which was red and green.
More colors can be easily added in order to have more colors and rings, then, the color
dependency problem can be solved. The size of the pipe that can be tested by using this
prototype is 1.85 inches or more, and by minimizing the size of the lenses the pipe size that
is needed to test can be kept to a minimum as well. The deformation can be clarified by
using eccentricity calculation. The frames that have defects is having eccentricity closer to
one, and the others are closer to zero as explained in chapter 5 when semi-major and semi-
minor axes have been calculated and used to determine the eccentricity by using the
following formula.
Misalignment problem also exists in this scanning and prevents the eccentricity calculation
from getting zero value for the areas that do not have defects. As a result, misalignment
correction had been applied to minimize this problem when the semi-major and semi-minor
axes also used to define the transformation matrix in order to squeeze the image to be closer
to circle.
63


6.2 Future Work
Optical sensors for NDE applications using structured light and 3D visualization is not a
small topic to cover in one thesis, but a lot of things should be considered to develop and
modify to get better results. Misalignment, prototype size, projection resolution, and image
reconstruction need more improvement and enhancement to be more productive.
For the prototype design, the number of colors and the size of the lenses can be enhanced.
More than two colors can be added to the slide. The lenses that needed to build the
prototype can be chosen smaller than 1.5 inches to scan the pipes that have a smaller
diameter. Likewise, a robot can be combined to prototype to avoid the misalignment
problem that occurs because of using human hands. The robot can be designed to move by
fixing steps and without tilting.
64


REFERENCES
[1] CCTVPipeline Testing Survey & Reporting with Wincan v8. Retrieved from
http://www.auscamqld.com.au/.(2013, Nov).
[2] Matzkanin, D. S. (n.d.). A Brief Introduction to Nondestructive Testing. The
AMMTIAC Quarterly,, 7, 7-10. Retrieved from http://ammtiac.alionscience.com
[3] TRAINING GUIDELINES IN NON-DESTRUCTIVE TESTING TECHNIQUES.
(2013). Retrieved from http://www.iaea.org/books
[4] Liquid Penetrant and Magnetic Particle Testing at Level 2. Vienna, Austria:
INTERNATIONAL ATOMIC ENERGY AGENCY. (2000).
[5] https://www.tec-eurolab.com/eu-it/esame-visivo-vt.aspx
[6] John Markanich, P. E. (n.d.). ELECTROMAGNETIC TESTING TECHNIQUES FOR
QUALITY GRADING OF NONFERROUS ROD AND WIRE. Pittsburgh, PA:
Foerster Instruments, Inc.
[7] Introduction to Non-Destructive Testing Techniques. Retrieved from
http://www.eis.hu.edu.jo/ACUploads/10526/Ultrasonic%20Testing.pdf
[8] Introduction to Non-Destructive Testing Techniques, (n.d.). Liquid Penetrant Testing.
Retrieved from http://www.eis.hu.edu.jo/ACUploads
[9] Introduction to Non-Destructive Testing Techniques, (n.d.). Magnetic Particle
Testing. Retrieved from
https://eis.hu.edu.jo/upload/38000000/Magnetic%20Particle%20Testing.pdf
[10] Gao L, Zai F, Su S, Wang H, Chen P, Liu L. Study and Application of Acoustic
Emission Testing in Fault Diagnosis of Low-Speed Heavy-Duty Gears. Sensors. 2011;
11(1):599-611.
[11] https://www.ndeed.org/EducationResources/CommunityCollege/Other%20Methods/
AE/AE_Intro.php
[12] https://www.infraredheaters.com/fundamen.html
[13] N. Pears et al. (eds.), 3D Imaging, Analysis and Applications,
DOI 10.1007/978-l-4471-4063-4_2, Springer-Verlag London 2012
[14] J-Angelo Beraldin, F. B. Active 3D sensing. Ottawa, Ontario, Canada: Institute for
Information Technology. (2000).
[15] Liu, Y. D. Electromagnetic Imaging Methods for Nondestructive Evaluation
Applications. 11, 11774-11808. doi: 10.3390/sl 11211774. (2011)
[16] Curless, B. (1999). Applications of Computer Vision to Computer Graphics.
Retrieved from ACM SIGGRAPH: https://www.siggraph.org/publications/
newsl etter/v3 3 n4/contributi ons/curl ess. html
65


[17] Shengyong Chen, Y. F. Active Sensor Planning for Multiview Vision Tasks (1 ed.).
Springer-Verlag Berlin Heidelberg, doi: 10.1007/978-3-540-77072-5(2008).
[18] Mi quel Massot, G. O. Optical Sensors and Methods for Underwater 3D
Reconstruction, pp. 31525-31557. doi:10.3390/sl51229864. (2015).
[19] Li, L. Time-of-Flight Camera An Introduction. Retrieved from
http://www.ti.com/3dtof. (2014).
[20] Pierre Olivier, V. P. LEDDAR OPTICAL TIME-OF-FLIGHT SENSING
TECHNOLOGY: A NEW APPROACH TO DETECTION AND RANGING.
WPlOl-vl.3. (2015).
[21] Light Detection and Ranging (LiDAR). (n.d.). Portland State University. Retrieved
from http://web.pdx.edu/~jduh/courses/geog493fl2/Week04.pdf
[22] LASER SCANNING INFRASTRUCTURE ASSETS: NEW CAPACITIES, NEW
OPPORTUNITIES. Autodesk Infrastructure Solutions. Retrieved from
https://www.cansel.ca/images/About-us/White_Papers/autodesk_whitepaper-
lidar%20scanning%20techniques%20and%20roi.pdf. (2012).
[23] Sanjeev Kumar, P. K. (IEEE). An optical triangulation methodfor non-contact
profile measurement. HEFEI UNIVERSITY OF TECHNOLOGY, doi: 1-4244-
0726-5/06. 2006
[24] Gerig, G. Structured Lighting. Retrieved from
http://www.cs.cmu.edu/afs/cs/academic/class/15385-s06/lectures/ppts/lec-17.ppt.
(2012).
[25] OConnor, C. The Nature Of Polyethylene Pipe Failure. 239 Retrieved from
https://pgj online. com/2012/12/27/the-nature-of-polyethylene-pipe-failure/(2012,
December).
66


Full Text

PAGE 1

OPTICAL SENSORS DEVELOPMENT FOR NDE APPLICATIONS USING STRUC TURED LIGHT AND 3D VISUALIZATION By ABDUSAMEA ELFAID B.S., Aljabel Algharbi University, 2010 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in parti al fulfillment of the requirements for the degree of Master of Science Electrical Engineering 2016

PAGE 2

ii 2016 ABDUSAMEA ELFAID ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Master of Science degree by Abdusamea Elfaid has been approved for the Electrical Engineering Program By Yiming Deng Chair Jan Bialasiewicz Chao Liu April 28 2016

PAGE 4

iv Elfaid Abdusamea. (M.S., Electrical Engineering) Optical Sensors Development for NDE Applications Using Structured Light and 3D Accelerating Visual ization. Thesis directed by Assistant Professor Yiming Deng ABSTRACT As pipeline infrastructure systems continue to age and deteriorate, efficient and effective repair or replacement, and maintenance scheduling to reduce the associated significant costs are critical but remain challenging. Pipeline inspection technologies and innovative solutions need to be improved and/or developed in order to provide cost effective solutions to assist the pipeline safety and reliability full decision support system. Mo reover, identification and classification of current vintage pipeline inner wall damage precursor are of critical importance. However, there are limited success in sensing and characterizing the small diameters of pipelines with high probability of detecti on (POD) [3] and the capability of the currently available accelerometers and imaging technologies that can be miniaturized and integrated into smaller size pipes for a fast scan is questionable and needs a systematic assessment. The major drivers of prem ature failure due to slow crack growth are bending stresses due to tight bend radii, impingement and fittings. Damage is also introduced by pipe squeeze off during maintenance operations. These conditions identification has been investigated by various non destructive evaluation (NDE) [2] techniques, such as direct visual/optical methods using CCD cameras, ultrasonic testing, liquid coupled acoustic measurement (e.g. sonar) and laser based surface inspection approaches including light detection and ranging ( L iDAR) [18] [19] and laser topography However, these currently available technologies suffer either from low sensitivity and

PAGE 5

v resolution for small damage precursors, or complex settings and large system footprint that makes it incapable for plastic pipes wit h much smaller diameters. In this thesis three different generations have been designed in order to obtain 3D visualization. The first generation concentrated to scanning the 3 inches pipes with linear defects. Laser source and camera with small view angl e have been used to collect the data. There are missing parts in the final 3D image because the angle of the camera is small to capture all the scene. For this reason, the second generation is illustrated to decrease the shadow area that occurs from the vi ew angle of the camera. Another camera with large view angle (fisheye camera) is used with the laser source to obtain full ring without missing parts. The third generation is structured light scanning. A new slide projector with two rings two colors has be en designed to capture the scene in order to generate 3D reconstructed image. A pipe with squeezed part is used to do this test by using the structured light scanning. 3D reconstructed image illustrated the squeezed part in the pipe, and this twisted part has a shape of an ellipse instead of a circle. Then, the semi major and semi minor axes are calculated to define the eccentricity in order to clarify the frames with squeezed part. Misalignment correction is applied to reduce the camera tilting that occurs from using human hands while doing the scan. The form and content of this abstract are approved. I recommend its publication. Approved by : Yiming Deng

PAGE 6

vi ACKNOWLEDGEMENTS I appreciate the inspiration, guidance, and assistance of many people without who m this work would not have been possible. Professor Yiming Deng my thesis advisor, provided guidance and support not only for this thesis, but also in my professional development. I would like to thank the Libyan government, especially the Ministry of Hi gher Education and CBIE for nominating me to study aboard and supporting me and my family with a fully funded scholarship. Finally, I must take the opportunity to acknowledge and thank my parents for their endless love and support. They have implanted in m e high work ethics and confidence to prosper

PAGE 7

vii TABLE OF CONTENTS Chapter 1. Introduction ................................ ................................ ................................ .................... 1 1.1 Background ................................ ................................ ................................ ........................... 1 1.2 Research Significance ................................ ................................ ................................ ........... 2 1.3 Thesis Outline ................................ ................................ ................................ ....................... 3 2. Nondestructive Evaluation (NDE) ................................ ................................ .................. 5 2.1 Introduction ................................ ................................ ................................ .......................... 5 2.2 NDE Methods ................................ ................................ ................................ ....................... 7 2.2.1 Visual and Optical Tes ting (VT) ................................ ................................ .................. 7 2.2.2 Radiographic Testing (RT) ................................ ................................ ............................... 8 2.2.3 Electromagnetic Testing (ET) ................................ ................................ ...................... 9 2.2.4 Ultrasonic Testing (UT) ................................ ................................ ............................. 11 2.2.5 Liquid penetrant Testing (PT) ................................ ................................ .................... 12 2.2.6 Magnetic Particle Testing (MT) ................................ ................................ ................. 14 2.2.7 Acoustic Emission Testing (AE) ................................ ................................ ................ 15 2.2.8 Infrared and Thermal Testing (IR) ................................ ................................ ............. 15 3 Optical Sensors ................................ ................................ ................................ ............ 17 3.1 Introduction ................................ ................................ ................................ ........................ 17 3.2 Passive Acquisition ................................ ................................ ................................ ............. 17 3.2.1 Multiple Viewpoints ................................ ................................ ................................ ... 18 3.2.2 Single Viewpoints ................................ ................................ ................................ ...... 19 3.3 Active Acquisition ................................ ................................ ................................ .............. 20 3.3.1 Active Stereo ................................ ................................ ................................ .............. 20

PAGE 8

viii 3.3.1.1 Laser Scanning ................................ ................................ ................................ ... 21 3.3.1.2 Structured Light ................................ ................................ ................................ .. 22 3.3.2 Triangulation ................................ ................................ ................................ .............. 26 3.3.3 Time of Flight ................................ ................................ ................................ ............ 29 3.3.3.1 Direct Time of Flight ................................ ................................ ....................... 29 3.3.3.1.1 LiDAR ................................ ................................ ................................ ........ 29 3.3.3.2 Range Gated Imaging ................................ ................................ ........................ 30 3.3.3.3 Phase Differenc e Measurement ................................ ................................ .......... 31 4. 3D Pipe Inner Wall Reconstruction ................................ ................................ .............. 33 4.1. General Overview ................................ ................................ ................................ .............. 33 4.2 Materials and P reparation ................................ ................................ ................................ ... 33 4.2.1 The D efects and C racks in the P ipes ................................ ................................ .......... 33 4.2.2 Scanning w ithout L ight S ource ................................ ................................ .................. 35 4.3. Single Ring Scanning ................................ ................................ ................................ ........ 37 4.3.1 Scanning with S mall V iew A ngle C amera ................................ ................................ 38 4.3.2 Scanning with Large View Angle Camera ................................ ................................ 41 4.3.3 Scanner and the Sensor in Opposite Direction ................................ ........................... 44 4.4 Multi R ings Mu lti C olors Scanning ................................ ................................ ................... 45 4.4.1 Projector Prototype ................................ ................................ ................................ .......... 46 5. Discussion and Analysis ................................ ................................ ............................... 52 5.1 Deformation Detection ................................ ................................ ................................ ....... 52 5.1.1 Eccentricity Calculation ................................ ................................ ............................. 52 5.1.1.1 Eccentricity of the Area with Free Crack Def ormation ................................ ...... 54 5.1.1.2 Eccentricity of the Area with S queezed P arts ................................ .................... 55

PAGE 9

ix 5 .2 Misalignment C orrection ................................ ................................ ................................ .... 58 6. Conclusion and Future Work ................................ ................................ ........................ 62 6.1 Conclusion ................................ ................................ ................................ .......................... 62 6.2 Future Work ................................ ................................ ................................ ..................... 64 References

PAGE 10

x LIST OF FIGURES F igure 1.1 a Illustration of ductile failure of PE pipe b Typical slit type fractures [21] .............. 2 2.1 sample that shows the destructive evaluation will affect the characteristic of the spacemen [4] ................................ ................................ ................................ ....................... 5 2.2 Illustration of the old technique of using visual testing [5] ................................ .......... 8 2.3 discontinuity detection by radiographic testing (x ray) [5] ................................ ......... 9 2.4 The principle of electromagnetic testing [6] ................................ .............................. 10 2.5 Illustration of using ferrous inclusion to detect the metal [6] ................................ .... 10 2.6 Illustration of main steps of Ultrasonic Testing [7] ................................ ................... 11 2.7 illustration of five steps of dete cting the flaws by using liquid penetrant testing [3] 13 2.8 Crack detection by using Magnetic particles testing [9] ................................ ............. 14 2.9 Illustration of Acoustic Emission detection [11] ................................ ........................ 15 2.10 Electromagnetic spectrum [12] ................................ ................................ ................ 16 3.1 Optical sensor techniques ................................ ................................ ........................... 17 3.2 Illustration of multiple viewpoints technique [13] ................................ ...................... 18 3.3 Illustration of generating 3D image by using shape from shading [13] ..................... 20 3.4 Illustration of active stereo with laser scanning technique [14] ................................ 21 3.5 Illustration of using laser and camera for detect the surface [16] ............................... 22 3.6 Illustration for some binary pattern [18] ................................ ................................ ..... 23 3.7 Illustration of the camera and the projector coordinates [17] ................................ ..... 25 3.8 Illustration of 2D triangulation ................................ ................................ ................... 26 3.9 Illustration of 3D triangulation [18] ................................ ................................ ............ 28 3.10 LiDAR scanning by using airplane [21] ................................ ................................ ... 31

PAGE 11

xi 3.11 Illustration of phase shifting [23] ................................ ................................ .............. 32 3.12 Illustration of calculating the Q 1 and Q 2 [19] ................................ ........................... 32 4.1 The pipes that have been used for the experimental work ................................ .......... 33 4.2 The holes that have been introduced as defects ................................ .......................... 34 4.3 The linear cracks in the pipes ................................ ................................ ...................... 34 4.4 Pipe with a small diameter (1.85 inches) and damage section ................................ ... 35 4.5 Optical inner view of the E41 damage of GTI pipe sample ................................ ....... 35 4.6 Two different cameras installed inside the pipe ................................ ......................... 36 4.7 The different defect types that have been captured by using the camera ................... 37 4.8 The laser source used in the sensor p rototype ................................ ............................ 38 4.9 Simple camera and source bundle prototype ................................ .............................. 38 4.10 Some frames form the collected data: (a) with no defects and (b) with defects ....... 39 4.11 Results aft er applying the thresholding to image frames shown in figure 4.10 ........ 40 4.12 Initial imaging result of 3D reconstruction using a stack of frames ......................... 40 4.13 Fish eye camera ................................ ................................ ................................ ........ 41 4.14 the camera and the laser source installed in temporary prototype ............................ 42 4.15 Complete ring images: a ring with some deformation. b Full ring. ....................... 42 4.16 The 3D reconstructed image with comp lete FOV ................................ .................... 43 4.17 The prototype that have been used in this section ................................ .................... 44 4.18 Illustration of one frame that has a defects ................................ ............................... 44 4.19 3D reconstruction ................................ ................................ ................................ ..... 45 4.20 structured light with using a projector to generate the pattern [24] .......................... 46 4.21 The lenses stacked inside the prototype ................................ ................................ .... 47 4.22 The slide for the projector ................................ ................................ ......................... 47

PAGE 12

xii 4.23 The projection on the rough surface ................................ ................................ ......... 48 4.24 Final shape of the projector ................................ ................................ ....................... 49 4.25 Projector and the camera connected to each other ................................ .................... 49 4.26. a th e projection on the smooth surface. b the prototype with blocked edges ......... 49 4.27. a Illustration of one frame without defects. b I llustration of a frame with squeezed part ................................ ................................ ................................ ................................ .... 50 4.28 Illumination of the white color has been reduced ................................ ..................... 50 4.29 3D reconstructed image with and without the defects. ................................ ............. 51 4.30 Some frames that show the squeezed parts inside the pipe ................................ ...... 51 5.1 Illustration of the circle which has fixed radius ................................ .......................... 53 5.2 Illustration of the ellipse [25] ................................ ................................ ...................... 54 5.3 Illustration of 20 frames from collected data ................................ .............................. 55 5.4 Illustr ation of one frame with a defect and there are two lines to emphasize the ring center ................................ ................................ ................................ ................................ 56 5.5 Final result of the eccentricity has been calculated for 100 frames ............................ 57 5.6 Illustration of the semi m ajor radius of all 100 frames ................................ ............... 57 5.7 the region without defects but the shape is not circle ................................ ................. 58 5.8 The eccentricity of multi frames without misalignment correction ............................ 59 5.9 3D reconstructed image of number of frames without applying misalignment correction ................................ ................................ ................................ .......................... 59 5.10 a illustration of one frame that effected by misalignment and the shape is not a circle. b The same frame after applying mi salignment correction and the shape is closer to circle ................................ ................................ ................................ .............................. 60 5.11 3D reconstructed image after applying misalignment correction ............................. 60 5.12 The eccentricity after applying misalignment correction ................................ ......... 61

PAGE 13

1 1. Introduction 1.1 Background There are a lot of developments in the last decades in infrastructures which made the distribution of plastic gas pipes more complicated and widespread across the US. Plastic pipes have some benefits t hat make this kind of pipes more sufficient than other types of pipes. Corrosion and chemical resistance of plastic pipes are very high because the conductivity of the plastic is minuscule and the electrolytic erosion has no impact. Moreover, these pipes c an be used to handle chemical solutions more than the other pipes, so, plastic pipes can live for a long term. In addition to these benefits, the friction loss is little because the inner surface of the plastic pipes is smooth then small power is required to transmit the fluid. Flexibility also is one of the benefits when to compare it with metal pipes installation. However, some discontinuities occur to the plastic pipes which lead to prevent providing reliable service. These d iscontinuities as a followi ng ; Manufacturing failures happen when the pipe is not manufactured precisely and lead to having a crack or squeezed surfaces. Construction problems: when the pipes not installed properly in the groun d then the pipe will be pressed. Tensile and external pr essure: external pressure happens when the pipe is buried in the ground and not symmetric load will be applied then the pipes will have some bending. Tensile failure has four types which are [ 2 1] Ductile tensile, Brittle tensile, Fatigue failure, and Compr ession failure Because of these failures, diagnostic testing should be provided to inspect the failures and give a sufficient discretion of the defects. There are two types of testing that can be used to i nspect these defects which are; Destructive Testi ng and Non destructive Testing

PAGE 14

2 Destructive testing mainly used to inspect the materials but will impact the characteristics of these materials as will be described in chapter 2 then the tested materials will not be useful. Therefore, Non destructive t esting have chosen to use in this thesis. NDE have a different type of methods which will be clarified in chapter 2 as well. Visual and Optical Testing is the method that used in this thesis. Laser scanning and structured light scanning are the primary met hods of new visual and optical testing. Laser scanning has been used to inspect the deformation of the inner side of the pipes [2]. In this thesis, pipes with 3 inches will be tested by using laser scanning t o generate a 3D image. A laser ring will concent rate inside the pipe then a camera will capture the view and collect the frames to process them by using a computer software. Moreover, structured light with multi colors multi rings will scan the small pipe with 1.8 inches to generate 3D reconstructed ima ge. Figure 1 1 a Illustration of d uctile failure of PE pipe b Typical slit type fractures [ 2 5 ] 1.2 Research Significance This thesis is focusing on inspecting the inner surface of two different type of pipes. Laser and structured light t echniques have been used to investigate and analyze the deformation of the collected data. Structured light biased on endoscope technique [4] will be used to test 1.8 inches. The principle of using structured light is to send a pattern of light toward a b

PAGE 15

3 the object then the illumination will be observed by using a camera. However, the main contributions of this thesis can be as a following: Making some deformation in 3 inches pipes to test them by using laser scanning technique. This step has three main catego ries which are: Scanning b y using small view angle camera, Scanning by using lar ge view angle (fisheye camera), and Generating 3D reconstructed image In addition, multi rings multi colors structured light will be explained to scan the pipes that have a di ameter 1.8 inches. And a new slide projector will be designed as well to generate multi colo rs multi rings structured light. Then, 3D reconstructed image will be generated. Finally, by calculating the eccentricity of each frame, the deformation can be exp lained in other way to show the defects without using 3D visualization. The eccentricity also can be used to reduce the misalignment that produced from the camera misalignment. 1.3 Thesis Outline Chapter Two: the difference between destructive and nonde structive techniques are discussed as well as the illustration of the different kinds of the NDE methods. Chapter Three: this chapter has the principles of optical sensors techniques which are divided into passive and active techniques. These methods will be discussed briefly including to the laser and structured light scanning. Chapter Four: All the experimental works that have been done in this thesis will be presented as well as the 3D reconstruction of each generation. Chapter Five: The deformation and the misalignment will be analyzed and discussed by calculating the eccentricity.

PAGE 16

4 Chapter Six: All the work will be concluded in this chapter in addition to the recommended future work.

PAGE 17

5 2. Nondestructive Evaluation (NDE) 2.1 Introduction Destructi ve testing also is known as mechanical test [1]. When some materials have tested by destructive testing methods, the materials will be destroyed, or the characteristic of these materials will be changed as illustrated in figure 2.1. Destructive testing can be Static or dynamic, and this test is useful to evaluate: Yield point, Fatigue life, Hardness, Ductility, Ultimate tensile strength, Elongation Characteristics, Impact Resistance, Toughness, and Corrosion Resistance as illustrates in [1] Although destru ctive testing sometimes gives a proper result, the materials that have been tested will not be useful anymore in same purpose [1] [2]. Figure 2 1 sample that shows the destructive evaluation will affect the characteristic of the spacemen [4] NDE is techniques that have been used to evaluate structures or materials without destroying or changing the system usefulness as destructive testing does [2] [3] Some other definitions have been used for nondestructive evaluation which are: Nondestructive Testing (NDT) or Nondestructive Inspection (NDI). In this chapter NDT, NDI, NDE will be used interchangeably.

PAGE 18

6 Destructive testing (evaluation) was the most known technique before NDE was invented. NDE is not just one technique that has the equal modality to detect the deformations, but different kind of techniques can be illustrated in various kind of obstacles [3] Visual and optical testing, Radiographic Testing, Electromagnetic testing, Ultrasonic testing, Liquid p enetrant testing, Magnetic particle testing, Acoustic emission testing, and Infrared and thermal testing are the NDE techniques that are used in different fields [2] NDE is not a method that solves and fixes the cracks or deformations in the tested materi als, but it helps to detect these defects in order to give right solutions that contribute to enhancing the efficiency of the materials. In the pipes companies especially the plastic pipes need to check the cracks and the production quality to detect wheth er if useful to handle the servers or need to have some maintenance. In this case, NDE methods should be provided to identify and evaluate the materials without destroying them [4] Likewise, in medical field NDE techniques are broadly used. For example: if a doctor needs to check the human gut, the endoscope will be passed inside the patient body to see the problems. Then the treatment will be prescribed according to the test result. Also, the X ray can be used to detect the sprain or the crack in the bo nes as the first step to give a proper explication. As a result, NDE is not a simple method to use in one area or in specific problems, but it can be utilized in any field because the idea of NDE is similar to what the human body use to sense the things i n their daily life. This type of the evaluation can be divided in to several partitions : NDE is used to study the materials properties as well as to their quality. The cost of the production and maintenance can be reduced when NDE methods are used. The time of the test can be reduced and the size of the materials can be measured [2] [3]. NDE will not give a right evaluation when the wrong method is used to check the materials. As a result,

PAGE 19

7 using a proper method of NDE will help to avoid or decrease the limi tation that may happen if the wrong method is applied. 2.2 NDE Methods 2.2.1 Visual and Optical T esting (VT) Visual and optical testing is the most common method that has been used for a while [1] [2]. Also is known as the easiest way to examine the materials in a ddition to being the initial step that implements before using any other NDE methods. The evaluation will lead to getting wrong results if VT has been used in not a proper reason. For example, if the material has some defects that are not showed on the sur face, the VT will not detect these defects because they located inside the material. Then by knowing the obstacle before choosing the method is the first step to getting the right evaluation [3]. VT is used to inspect the materials that have some failing i n such as corrosion, Holes, Cracks, and Blisters [1][3]. In this thesis VT is used in order to evaluate the cracks in PE and PVC pipes. For old VT method, simple types of equipment are used to detect the flaws as illustrated in figure 2.2; a just particul ar mirror can be sufficient to see the flaws. This simple method has a lot of limitations that make it not useful when complicated objects have been tested. For complicated objects, the human vision becomes harder to use [3]. Therefore, the optical and the computer vision should be used to evaluate this kind of failings. The new method that is used the computer vision [3] to evaluate the materials is more beneficial. The laser scanner and structure light are used to illuminate the object, and then a sensor is used to capture the scene in order to generate 3D visualization and evaluate the cracks in the materials. Laser scanning and structured light scanning will be discussed in details in chapter 3.

PAGE 20

8 Figure 2 2 Illustration of the old technique of using visual testing [5] 2.2.2 Radiographic T esting (RT) RT is also one of the NDE methods that are widely used. After Wilhelm Conrad Roentgen has discovered the X ray in 1895 [2], the NDE started using this techniq ue as a one of the primary methods to check the discontinuities of the materials [2] [3]. Figure 2.3 shows the principle of the radiographic testing (x ray). This method has been used in medical and engineering experiments. The most important feature of th is method is hidden flaws can be detected and displayed in real time. As some other methods, RT has advantages which are: The surface and the under surfaces can be scanned as well as hidden parts can be detected, and most of the materials can be scanned by this method. RT also has some limitations which have some health issue can be occurred because of the radiations that come from the X ray [3]. Besides, a good experience is required to handle this method as well as the expensive equipment that are needed [1] [3].

PAGE 21

9 Figure 2 3 discontinuity detection by radiographic testing (x ray) [5] 2.2.2 Electromagnetic T esting (ET) Electromagnetic testing methods have been used to detect the cracks and flaw in the nonferrou s materials such as rod and wire. Figure 2.4 shows tested material surrounded by the coil and alternated current is sent through this coil. Then the receiver coil detects the current that have interacted with the material. Finally, the received signal is compared with the reference signal to show the cracks [3]. In ET imaging, ET methods are essential in NDE area. Maxwell's equations are essential in all ET methods as well as covering a wide range of electromagnetic spectrum from static and direct current. ET methods are fundamental in order to detect the flaws in a case of dielectric material and conducting materials as illustrates in [ 15 ]. Microwave imaging is one of ET applications in addition to eddy current and terahertz imaging. In [ 15 ] Deng and Liu have divided the ET methods according to the wavelength (from long wavelength to short wavelength). These methods based on this classification

PAGE 22

10 are; static electromagnetic methods, Quasi Static imaging methods, and high frequency time varying imaging metho ds. Each technique has been explained in [ 22 ]. Figure 2 4 The principle of electromagnetic testing [6 ] Figure 2 5 Illustration of using ferrous inclusio n to detect the metal [6]

PAGE 23

11 2.2.4 Ultrasonic T esting (UT) Ultrasonic Testing is one of the methods that use high frequency, ultrasonic waves, and mechanical waves to evaluate the materials. UT has been applied in both medical and industrial inspection [2]. Cracks and flaws can be recognized by using high frequencies (0.5 to 15 MHz) [7]. The primary process of using the UT is to send a signal using transducer then if there is any crack or flaw in the object the signal will be reflected to the receiver with di fferent time [7] and magnitude as shown in figure 2.6. And then the signal will be displayed and analyzed. UT just is needed one side to do the ins pection ( all surface is not required to do the test). Also, UT has high accuracy as illustrated in [7]. UT do es not have health issues same as RT does. UT uses portable equipment which makes this method more useful. The surface and subsurface are sensitive to the test [3] [7]. UT is not beneficial for some materials according to sound transmission as well as it is not useful for small, rough, and not homogenous materials. UT needs reference signal to compare it with the received signal. Moreover, the surface should be accessible to transmit the ultrasound signal. And the last disadvantage is UT needs advanced training comparing with the other methods. Figure 2 6 Illustration of main steps of Ultrasonic Testing [7]

PAGE 24

12 2.2.5 Liquid penetrant T esting (PT) Liquid penetrant testing is one of oldest and simplest method in the NDE technique because this method does not use any new technology t o detect the flaws and cracks. The instructor just needs simple tools to evaluate the sample. But non porous feature should be existent to ensure the test is sufficient [8]. There are several steps to implement PT test which is started from cleaning the su rface from any kind of oils or dust. Then the penetrant material is applied to the sample. There are two types of the penetrant material which are: visible penetrant and fluorescent penetrant. After implementing the penetrant to the material, the dwell tim e should be considered before removing the penetrant. Next step is to eliminate the penetrant. The penetrant can be moved by water washable, post emulsifiable, lipophilic, solvent removable, or post emulsifiable, hydrophilic [3] [8]. The last step is to d etect the cracks and flaws by using the developer as shown in figure 2.7. This action takes some time to detect the cracks spatially for the materials that have small cracks. PT it is portable technique and the cost is low as well. The order advantages are PT can detect the fine cracks and different kind of materials can be identified by this method. This method as mentioned is one of the easiest methods to identify the flaws in some material. But comparing with the other NDE methods, PT still has limitati ons that make this method not sufficient in some cases. Such as just nonporous materials can be inspected, pre cleaning can hide some of the defects as illustrated in [8] flaws should be on the surface, inner flaws cannot be detected, and removing the pen etrant after doing the test is required.

PAGE 25

13 Figure 2 7 illustration of five steps of detecting the flaws by using liquid penetrant testing [3]

PAGE 26

14 2.2.6 Magnetic Particle T esting (MT) The basic concept of using MT m ethod is taking advantage of the magnetic field of the materials. MT is adopted to investigate a diversity of product forms containing forgings, castings, and weldments. Before implementing MT, the characteristics of the material should be identified as illustrated in [2]. Just ferromagnetic materials can be inspected by using this method. Moreover, the flaw should be on the surface of the tested material to be able to recognize. Otherwise, t his method will not be useful. When the material is magnetized the flux lines will travel from south to the north p ole. Since the material has a crack, the flux will exit the material and then reenter (flux leakage) again to the material as shown in figure 2.8 [9]. An iron particle is used to check the leakage. The Iron will be attracted by the flux leakage, and then the crack can be defined easily by knowing the attracted area. Figure 2 8 C rack detection by using Magnetic particles testing [9]

PAGE 27

15 2.2.7 Acoustic Emission T esting (AE) The fundamental principle of the acoustic emission is to use elastic waves. The acoustic emission source is used to sent the elast ic waves toward the material (that is needed to test) via a transmission media. The wavies change to the electrical signals in order to be magnified and processed. Then the defects can be displayed on th e screen as illustrated in [10] Figure 2 9 Illustration of Acoustic Emission detection [11] 2.2.8 Infrared and Thermal T esting (IR) Infrared and thermal testing is also known as infrared vision [2]. Infrared waves located in the electromagnetic spectrum as shown in the figure 2.10. For infrared waves, there are Mid Wave IR (MWIR) and Low Wave IR (LWIR) which both of them have a different wavelength. The wavelength of LMIR is between 7 and 14 m, but the wavelength of MWIR is between 3 and 5 m. By using the infrared se nsor to detect the surface. When thermal waves are applied on the material, the cracks and flows will interact with the waves differently comparing with areas that do not have defects. By measuring the difference of the thermal waves, the defects can be d etermined. Moreover,

PAGE 28

16 IR uses expensive equipment such as a thermal camera which makes this technique has a cost limitation comparing with efficient and cheaper methods [2]. Figure 2 10 Electromagneti c spectrum [12]

PAGE 29

17 3 Optical S ensor s 3.1 Introduction Optical sensing is one of the NDE techniques that have been improved significantly in past few years. There are several ways under optical sensing that produce a 3D image. However, two different t echniques will be the main categories which are passive acquisition and active acquisition, in which laser and structured light will be classified under active acquisition. The following chart is shown in figure 3.1 illustrates the main categories of the O ptical sensor. Figure 3 1 Optical sensor techniques 3.2 Passive Acquisition Passive Acquisition is one of the methods that can be used to get 3D optical images. This technique is called passive because the data is collected by using the camera without using an active source, such as lasers to provide strip line or by using structured lights. Multiple viewpoints and single viewpoint techniques are classified to be the primary methods for the passive acquisition.

PAGE 30

18 3.2.1 Multiple V iewpoints This technique is used to take the image from different viewpoints/angles. Two cameras or more can be used to collect the data from various angles. By using multiple cameras, the technique is called stereo vision. There are dif ferent kinds of stereo with a different number of the cameras that have been used. For example: if two cameras have been used to capture the image the technique is called binocular stereo. Similarly, trinocular stereo is when three cameras have been used. Also, if a single camera has been used to take the image at the same point but from different locations and time, this technique is called structure from motion. By processing the images that have been taken from the different locations or different camera s, the 3D rays will be determined. Finally, from the 3D rays, the 3D position of the point in the scene can be determined. One illustration of the multiple viewpoints is shown in figure 3.2 Figure 3 2 Ill ustration of multiple viewpoints technique [13]

PAGE 31

19 3.2.2 Single Viewp oints In this technique, the captured image does not come from multiple cameras or the camera motion. The captured image, in this case, is taken by using the object details, for example, th e texture of the object, shading, or focus, etc. Figure 3.3 shows that shape from shading has generated the 3D image. This technique depends on the reflection from the object. The pixels in the reconstructed image illustrate the intensity of the reflection that comes from the object shade, and by using regularized surface fitting, the 3D image can be reconstructed. Overall, the multiple viewpoints method is more accurate and more efficient nhancement of the reconstructed image are described in [1]. Shape from shading technique is known as Photometric stereo. The Main steps to generate a 3D image by using this technique is that taking more than one image at the same point but the illuminatio ns for the scenes are different. The other technique is called shape from focus. Taking two images from different depths of field is the main idea of processing 3D image, and the approach is described in [3]. In summary, single view technique is not as good as multiple viewpoints technique in terms of speed and regulation. Therefore, multiple viewpoints technique is more commonly used. However, the passive acquisition is not suitable for this project. Active acquisition will be the tool for making the 3D imaging which will be reviewed and discussed in the following section.

PAGE 32

20 Figure 3 3 Illustration of generating 3D image by using shape from shading [13] 3.3 Active Acquisition Active Acquisition has differen t techniques to capture the viewpoints comparing with passive techniques. As illustrated in the passive techniques, captured image does not need structured light or laser strip to collect the data. However, in Active Acquisition, optical detectors such as camera need to be used to detect the spot. Besides, the active sources such as laser source or structure light should be considered to complete the imaging process. 3.3.1 Active S tereo Active stereo has a similar idea with multiple viewpoints. In this tec hnique, a light source with unique features has been used to replace the function one of the cameras in the multi ple viewpoints method. Figure 3.4 illustrates the principle of the Active stereo method with laser scanning. The light source focuses on the sc ene, and then the camera captures the view. There are different kinds of the light sources. Structured light and the laser are two of the most known sources. These techniques will be discussed in the following sections. Active stereo is more common than pa ssive techniques because of the high

PAGE 33

21 resolution and the data that have been taken by the laser, or structured light are more accurate. However, to generate a 3D image by using a camera and light source, the triangulation should be considered to calculate t he depth of the scene, which makes the method mathemati cally rigorous and challenging. Figure 3 4 Illustration of active stereo with laser scanning technique [14] 3.3.1.1 Laser Scanning A lot of applications have used laser scanning to get a 3D image. Laser scanning uses a strip of laser light, or it can be a circle shaped source to scan the inner surface of an object, e.g. pipelines in this project, which will be discussed in details later. Basically, a camera with high resolution uses as a sensor to capture the scene. Both the camera and the laser source functionin g as one unit to produce the 3D image. Figure 3.5 illustrates how laser and camera work together. The distance between the camera and the laser source, and the illumination of the object s. When the laser

PAGE 34

22 projects the light on the object, the laser light will capture the object shape. Then by knowing the camera and laser location, the depth of the object can be easily defined. Figure 3 5 Illustration of using laser and camera for detect the surface [16] 3.3.1.2 Structured Light As discussed in laser scanning section, structured light is similar to laser technique in setup; however, the structured light projector has been used in this tec hnique instead of the laser source. The structured light technique has been utilized for a lot of applications, for example, industrial and medical testing. The principle of struct ured light is shown in Figure 3.6, which is depending on the relative positi ons of camera and projector. In our prototype, the projector is connected to the computer and will be controlled to generate pre designed patterns. There are a different kind of patterns that have been used for the structured light such as: Spatial pattern s, temporal patterns, Color patterns.

PAGE 35

23 Figure 3 6 Illustration for some binary pattern [18] When the projector projecting the light on the scene, the camera will capture the illumination from the object. Likewise, the light will take the object shape. By knowing the int ersect point between projector light and the camera view, the 3D image can be calculated. The 3D coordinates for the point as shown in the figure 3.7 is given by Then the image that have seen by the camera can be determined by the following: And the is the matrix that present the camera perspective

PAGE 36

24 Also for the projector, the 3D point in the object according to the projector coordinates Then back projection is going to be By capturing the image, the relation between the image that hav e been captured by the camera and the proj ector view has some factors such as; rotation and translation. Then the relation can be written as ) ) When are t he matrices for the rotation in different axis and the translation Then ) By multiply new matrix will be defined when 4D row vectors ) ) ) From equation (Eq3.1 ) and (Eq3.7 )

PAGE 37

25 ) If ) Then ) Finally the 3D position for any point in the object can be calculated b y knowing H matrix ) Figure 3 7 Illustration of the camera and the projector coordinates

PAGE 38

26 3.3.2 Triangulation Triangulation is an essential step to building a 3D image because the relation between the camera and the light source should be defined to calculate the depth. Triangu lation is used in the passive and active acquisition. However, in a following steps triangulation process of the active acquisition will be described. Figure 3.8 shows a 2D triangulation system [23]. And the approach is described in [24]. The distance bet ween the camera and the light view and the light source projection. L O P are the points that determ ine the location of P in the object. Distance d can be calculated by knowing the angles By using sine la w Figure 3 8 Illustration of 2D triangulation

PAGE 39

27 ) Then ) When ) ) Now the coordinates of the point P can be determined by the which is controlled by t he projection of the light source on the project. ) When ) ) Similarly, t o find 3D image points that described in [24] and illustrated in figure 3.9 the 3D coordinators can be determined by knowing and b From ray theorem ) ) Z can be defined by angle. 0 ) )

PAGE 40

28 ) ) ) ) Then the coordinates will be as following: ) Figure 3 9 Illustration of 3D triang ulation

PAGE 41

29 3.3.3 Time of Flight Laser and structured light use visible light that can be seen by human vision but there are other signals e.g. (invisible light) can also be useful for scanning. In the applications that use invisible light, the Time of Fligh t (TOF) is one of the representations that used to define the depth or the distance. TOF has been illustrated in [19] [20] and the signal that will send from the source is modulated, then this signal will be reflected toward the receiver. There are three m ain divisions of TOF. Each one has a different approach with the transmitted signal [20]. 3.3.3.1 Direct Time of Flight This technique uses discrete pulses to send a signal and receive it by a sensor. According to the time difference between the transmi tted and reflected signal, the distance can be calculated easily by using the following formula. ) When c is the light speed c= 299,792,458 m/s Direct time of flight is difficult to implement as illustrates in [20], when the accuracy is needed to be in centimeter level. High clock speed is required in order to increase the accuracy [20]. 3.3.3.1.1 Li DAR Light Detection and Ringing (LiDAR) has been developed since the 1960s [21]. LiDAR is one of the techniques that depends on TOF process. A beam of a laser is adopted as a source for LiDAR. This laser beam uses to measure t he distance between the object and the source.

PAGE 42

30 However, LiDAR is known as a fast scanner because this technique can collect the data more quickly than the other techniques. There are different kinds of the LiDAR some of them use fixed LiDAR to collect the data, and the others use moving LiDAR. Infrastructure engineering is one of the applications that dependence on the fixed LiDAR as illustrates in [20]. Airborne LiDAR, Mobile Terrestrial LiDAR, and Static Terrestrial LiDAR are the main types of LiDAR techn ique [21]. Airborne LiDAR is used for the applications that need to scan from particular high, and the others are used on the ground. As discussed in the TOF, the speed of light and the reflection time are required to calculate the distance. GPS is needed in order to define the X, Y positions [22]. Ther e fore, the accuracy of the LiDAR system depends on the GPS accuracy, not just the laser reflection. Figure 3.10 shows an image has taken by LiDAR scanner. This system has some significant equipment functioni ng together to generate the 3D image such as Laser scanner, GPS, Data storage and management systems, High precision clock, GPS ground station, and IMU Inertial navigation measurement unit [21]. The laser frequency that is applied in LiDAR devices is aro und 50 kHz to 200 kHz [21]. Also, the wavelength of the laser beam is different from the system to another. In Doppler LiDAR as discussed in [21] the wavelength between 1500 and 2000 nm (infrared). In contrast, terrestrial mapping the wavelength between 10 40 and 1060 nm (near infrared). 3.3.3.2 Range Gated Imaging Range gated Imaging is not same as direct time of flight in calculating the distance. Range gated Imaging as illustrates in [20] depends on signal integration in order to calculate the distance A CCD or CMOS are used to capture the energy then the distance can be measured according to the different intervals of the energy ratio. The dynamic range of the

PAGE 43

31 CCD and CMOS is limited [20]. Therefore, the strong light in the surrounded area will affec t the measurements. Figure 3 10 LiDAR scanning by using airplane [21] 3.3.3.3 Phase Difference Measurement The phase difference between transmitted and received signal can be used to calculate the distance as discussed in [20]. The transmitter sends a pulse as illustrates in figure 3.11 then this signal will be reflected back to the receiver. The reflected signal has a phase difference comparing with the original. The receiver has a feature that works to change the received pulses to electrical curren t as described in [19]. reflected signal and the receiver. Then the depth can be calculated by using following formula. )

PAGE 44

32 Figure 3 11 Illustration of phase shifting [23] When c is speed of the light are the electrical charge for pulse signal determined in same time as the fallowing figure Figure 3 12 Illustrati on of calculating the Q 1 and Q 2 [19]

PAGE 45

33 4. 3D Pipe Inner Wall R econstruction 4.1. General Overview As known NDE technique will be used in this research and specifically the Visual and optical testing method will b e considered a main method. Also, the structured light is considered as our light source in this step. Therefore, the designing of the new prototype that uses structured light with multi rings multi colors will be explained in this section 4.2 Materials a nd P reparation 4.2.1 The D efects and C racks in the P ipes In the beginning, 3 inches diameter pipes are consider ed as the specimen. Figure 4.1 shows the pipe that has been used in the experimental work. Figure 4 1 The pipes that have been used for the experimental work Then some defects have been made in the pipe as illustrated in figure 4.2. Through hole defects with different diameters were introduced. In the white pipe specimen, the smallest hole in th e pipe has the diameter of 0.0787 inches and the biggest hole is 0.314 inches. Also, there are additional holes with 0.157 and 0.197 inches diameters. Moreover, some

PAGE 46

34 screws have been installed to mimic diff erent defect types. Figure 4.3 shows the cracks th at have different directions in the pipe and the width of the crack is 0.039 inches. Figure 4 2 The holes that have been introduced as defects Figure 4 3 T he linear cracks in the pipes

PAGE 47

35 Figure 4 4 Pipe with a small diameter (1.85 inches) and damage section The inner view of the damage area is shown using commercially available optical camera in Figure 4. 5 Fi gure 4 5 Optical inner view of the E41 damage of GTI pipe sample 4.2.2 Scanning without L ight S ource Before doing the laser scanning or structured light scanning, passive scanning is considered to show the d efects without any light source. However, two different cameras that have been used to scan the inner side of the pipes as shown in figure 4. 6 These cameras have

PAGE 48

36 different diameters. The diameter of the smallest camera is 0.23 inches and the diameter of t he largest camera is 0.62 inches. The camera is installed in the special prototype that has been produced by using 3D printer in our lab. After installing the camera and moving it inside the pipe, a video will be recorded to capture all inner surfaces in t he pipe. Figure (4.7 a) illustrates the cracks that have been captured by using the cam era. Also, in the figure (4.7 b) shows the holes that have been introduced in the pipe. Figure 4 6 Two different ca meras installed inside the pipe

PAGE 49

37 Figure 4 7 The different defect types that have been captured by using the camera 4.3. Single R ing S canning In this part of the research, the single ring is consi dered to be the light source of the scanner. Laser source with specific red ring has been used to detect the cracks and the deformation as shown in figure 4. 8 The relation between the diameter of the ring and the distance between the ring and the laser so urce is (1:1). This means that if the distance between the laser and the scene is 1 meter then the diameter of the ring on the scene will be 1 meter.

PAGE 50

38 Figure 4 8 The laser source used in the sensor prototype 4.3.1 Scanning with S mall V iew A ngle C amera The camera that has been showed in the figure 4. 6 is used with the laser source to scan the pipe. Figure 4. 9 illustrates the pro totype that has been used for scanning. Actually, the camera in this prototype will be connected to the laser source and shifted few centimeters back. This shifting because the view angle of the camera is small. So some shifting is needed to make the camer a capture the rest of the laser ring. By moving the prototype with constant speed, the camera will collect the data and will be saved in the computer. Figure 4 9 Simple camera and source bundle prototype

PAGE 51

39 As shown in figure 4. 10 the laser ring is not completed because the camera will capture a part of the laser source. As a result, some of the scene will be blocked. This kind of missing is called shadowing. This shadow area blocks about 25% of the whole scene. So this kind of missing is one of the limitations that should be considered to solve in the next sections. However, the laser ring will take the inner surface shape of the pipe. Figure (4. 10 a) illustrates that the laser ring does not has any deformation. On the other hand, figure (4. 10 b) shows some deformation that happened to the laser beam. The changes in the laser light shape will be considered as indications of defects that we want to s ee in the reconstructed images. (a) (b) Figure 4 10 Some frames form the collected data: (a) with no defects and (b) with defects Then the collected data were processed by recon algorithms. Some of image processing tasks will be considered to generate the final 3D image. In the beginning, thresholding and some digital filtration will be applied, then the result will be shown in figure 4. 11 Moreover, digital image enhancement techniques have been used to enhance and detect the deformation such as image closing, dilation, and other morphological process. Finally, after repeating this process for all the frames, the 3D image reconstructed can b e easily

PAGE 52

40 generated. Fig ure 4. 12 shows the 3D reconstructed image. There are some holes in the reconstructed image that show the defects that already have been done in the first section. Clearly some of the pipe will not be reconstructed because of the sha dows that have been done by the laser source. As result, the next section will discuss how we can get full 3D reconstructed image. Figure 4 11 Results after applying the thresholding to image frames shown in figure 4.10 Figure 4 12 Initial imaging result of 3D reconstruction using a stack of frames

PAGE 53

41 4.3.2 Scannin g w ith Large V iew A ngle C amera In this section we will repeat the same 3D reconstruction but with different prototype. Actually, the missing part in the previous 3D image will cause some problems. For example, if there are any defects or cracks in the miss ing area, the 3D reconstructed image will not show this defect. So, some enhancement should be done t o avoid this kind of problems. Using large view angle camera is one of the solutions that make the reconstructed image better. The difference between thi s camera and the previous one is the view angle. In the old one the view angle is about 45 degree, but in the new one is 180 degree that makes the whole scene can be detected by putting the camera and the laser source in the same base line. The new camera is called fisheye camera. Convex lenses have been installed to this camera to make the view angle bigger. Figure 4. 13 shows the fisheye camera that has been used in this experimental work. Figure 4 13 Fish eye camera In the last prototype the camera was contacted to the laser source but in the new prototype the camera and the laser source will be disconnected and they will be in same base line. Because the camera has 180 degree view angles, all the ring will be captured without

PAGE 54

42 missing any part. The new prototype can be illustrated in the figure 4. 14 Then, the prototype will be passed in to the specimen to collect the data. However, the scanning will be by human hand So, some misalignment will be considered. After that, the same 3D reconstruction process will be repeated. Then by taking a few frames from the recorded video, the laser light inside the pipe will be as shown in figure (4. 15 a b). In the figure (4. 15 a) the ring is completed and the laser light does not have any deformation. This indicates that the areas that have been covered by this light are still without any defects. On the other hand, figure (4. 15 b) shows the small d eformation in the laser light. Figure 4 14 the camera and the laser source installed in temporary prototype Figure 4 15 Complete ring images: a ring with some deformation. b Full ring.

PAGE 55

43 According to this deformat ion, this frame will represent the area that have the defect in the pipe. Likewise, by stacking all the frames and generate the 3D image, the defects will be clearly appeared in the final image figure 4.16 Figure 4 16 The 3D reconstructed image with complete FOV Actually, the laser and camera are taking a large space inside the pipe that make the minimization of the prototype more complicated. Different prototypes are being developed to minimize the size o f the sensor and will be discussed in next section

PAGE 56

44 4.3.3 Scanner and the S ensor in O pposite D irection As mentioned in the previous section that the laser source and the camera will take more space if they have been installed in the same baseline. Ther efore, in this section different prototype will be illustrated. Although this prototype will reduce the size of the scanner, the image will not be full ring. Some missing part will be shown in the collected data because of the metal that has been used to c onnect the laser source and the camera. A part of the light will be blocked and will not be recorded by connecting the prototype as shown in figure 4.17. Therefore, when moving the prototype inside the pipe, the scene will be as shown in figure 4.18. Fig ure 4 17 T he prototype that have been used in this section Figure 4 18 Illustration of one frame that has a defects

PAGE 57

45 Clearly, the defects can be illustrated a s missing p o ints in the laser beam. As a result. After processing all the frames the final 3D reconstructed image will be as s hown in figure 4. 19 Figure 4 19 3D reconstruction 4. 4 M ulti R ings M ulti C ol ors S canning As has been discussed in chapter 3, structured light is one of the methods that has been used to generate a 3D image by using the active technique. To produce a pattern that is needed to illuminate the scene, a projector is required to do thi s step. In typical 3D imaging by using structured light, the projector will be similar to one that is shown in figure 4.20. However, this kind of projectors will not be sufficient inside the small pipes. As a result,

PAGE 58

46 another device is needed to do this wor k in order to fit inside the pipe. A small slide projector has been developed to provide the particular pattern. Figure 4 20 structured light with using a proj ector to generate the pattern [24 ] 4. 4 .1 Proje ctor Prototype The main idea of using the slide projector is to emit a pattern with different colors on the object to obtain different illumination areas. As a result, a small slide projector has been designed to generate the colors instead of using the re gular projector. Moreover, to create this projector, some lenses are needed. The lenses that have been chosen for this step have a diameter (1.5 inches). Therefore, the size of the projector will go for the small pipes that we need to scan. Then, these len ses will be stacked with a certain way inside a prototype that has been de signed by using the 3D printer. One achromatic lens, two double convex lenses, and two convex lenses are used as shown in figure 4.21. Every lens has a focal length that should be co nsidered when the designer stacks the lenses together. The main idea of using the Achromatic lens is to reduce effects of chromatic and spherical aberration. The other lenses are used to magnify the image and increase the resolution. The slide that is used in this prototype has multiple rings as illustrates in the figure 4.22. Moreover, the

PAGE 59

47 power of the light source is strong enough to generate the light that is needed to emit the pattern. Figure 4 21 Th e lenses stacked inside the prototype Figure 4 22 The slide for the projector However, the slide has a two different colors (red and green) as shown in figure 4.22. There is a black area between the two ring s just to separate the edges. This separation can be removed in the future after increasing the resolution. Figure 4.23 shows the pattern on the rough surface to clarify the edges.

PAGE 60

48 The length of the projector is 7.87 inches, and the largest diameter is 1. 77 inches which can be used to scan the pipe that has 1.85 inches. Also, the distance between the light source and the rings when are projected inside the pipe is 1.02 inches. Then the next step is to connect the prototype and the camera as shown in figure 4.25. The camera that has been used in this step is the one that has small view angle. Because the pipe has a small diameter, small view angle camera can be used instead to the large view angle camera. As shown in the figure the distance between the light source and the camera is 5.51 inches. Figure 4 23 The projection on the rough surface By moving the prototype inside the pipe, the data will be collected by using the camera. Each frame has some features that came from the structure of the pipe. Figure 4.27 (a) illustr ates some frames that have been taken from the data. Clearly, the shape of the rings has been changed as shown in figure 4.27 (b). This changing shows the squeezed parts in the pipe. As a result, the reconstructed image will have the same shape of the corr upted image. But in this case, there is some illumination that has been generated from the extra light that comes from the lenses. To block the extra light, a black tape has been used to cover the edge of the lens as shown in figure 4.26 (b).

PAGE 61

49 The differ ence can be easily illustrated in figure 4.28, and the colors clearer after blocking the white light. Then, the algorithm will process each frame in order to generate the 3D reconstructed image as shown figure 4.29. Figure 4 24 Final shape of the projector Figure 4 25 Projector and the camera connected to each other Figure 4 26 a the projection on the smooth surface. b the prototype with blocked edges

PAGE 62

50 Figure 4 27 a Illustration of o ne frame without defects. b Illustration of a frame with squeezed part Figure 4 28 Illumination of the white color has been reduced A color segmentation is done as a first step in the algorithm to generate the 3D image. The red color is segmented and reconstructed as shown in figure 4.29. Also. Can be clearly seen the on e of the reconstructed images is not regular ring shape but it is twisted. The twisted area in the image illustrates the squeezed part inside the pipe. Moreover, figure 4.30 shows just a few number of frames that show the pressed area.

PAGE 63

51 Figure 4 29 3D reconstructed image with and without the defects. Figure 4 30 Some frames that show the squeezed parts inside the pipe 400 mm 0 mm 199 mm 357 mm

PAGE 64

52 5. D iscuss ion and Analysis 5.1 D eformation D etection 3 D visualization is an efficient method to emphasize the deformation in the pipes. But in order to clarify the deformation by checking the frames if in a circle shape or not, the eccentricity can be calculated to test the circular ity. In this chapter, the eccentricity will be calculated, and the misalignment correction also will be provided in order to minimize the camera misalignment. 5.1.1 Eccentricity C alculation As known the circle has a fixed diameter or radius in all direc tions as shown in figure 5.1. This radius is measured from the center of the circle to the edge. Therefore, the distance (a) equals to distance (b). On the other hand, in the case of the ellipse, the (a) and (b) are not equal. The ellipse has two different radiuses which are called semi major and semi minor. Semi major denoted as (a) and indicates to the larger radius, and the other radius (semi minor) denoted as (b). As seen in figure 5.2 there are two focal points which F1 and F2, and the distance between the focal point and the center is called linear eccentricity. Then by choosing any point on the ellipse (p) and calculate the distance between it and the focal points we can get the semi major axes as following: And Then by knowing (a), (b) and (f) other term can be calculated which is called eccentricity and denoted by (e). In case of ellipse, ( e ) will be between 0 and 1 (0 < e > 1).

PAGE 65

53 For example if we have a=2.5 and b=2, then Which is less than 1 Also if a=b=2.5 Which is in case of circle As a result, by calculating the eccentricity for each frame, we can define if the frame in shape of circle or not. Figure 5 1 Illustration of the circle which has fixed radius

PAGE 66

54 Figure 5 2 Illustration of the ellipse [25] 5.1.1.1 Eccentricity of the A rea with F ree C rack D eformation In this section, the eccentricity form ula will be applied for some frames that do not have a deformation to emphasize whether if it has a circle shape or not. Because of some other effects, the eccentricity may not be 0 in the case of free deformation area. This issue is called misalignment an misalignment is negligible, and just focusing on the large variation of the sequence. Then the defects can be easily illustrated. Figure 5.3 illustrates the eccentricity of 20 frames with out defects and the variation is not significant. The eccentricity in figure 5.3 varies between 0.26 and 0.32. By taking a special case to check the eccentricity. And frame number 60 has been chosen then the semi major radius is 200 (number of pixels). An d semi minor radius is 193 pixels then

PAGE 67

55 Figure 5 3 Illustration of 20 frames (4.4 cm) from collected data It is still not zero but this result will be compared with the results in the following sec tion 5.1.1. 2 Eccent ricit y of the A rea with S queezed P arts For the frames that have defects, the eccentricity will be much bigger than the other that have illustrated in the previous section. Therefore, by picking some frames that have say ten frames (120 130). One of the frames is shown in figure 5.4. Vertical and horizontal lines are added to the image to illustrate the center of the ring. However, the eccentricity will be calculated for frame number 125. Then by calculating semi majo r radius which is 217 and the semi minor radius will be 185

PAGE 68

56 The eccentricity has been increased comparing with the previous result. To emphasize the difference between the free crack area and squeezed part the eccentricity will be calculated for all the frames as shown in figure 5.5. In the beginning, the eccentricity will be fluctuating around small value then suddenly will start increasing until reaching the pick value (0.58) after that will decrease until start fluctuating around small value again. From this values, the deformation can be easily defined by the eccentricity. Figure 5 4 Illustration of one frame with a defect and there are two lines to emphasize the ring center

PAGE 69

57 Figure 5 5 Final result of the eccentri city has been calculated for 71 fram es (17.6 cm) Figure 5 6 Illustration of the semi major radius of all 71 frames (17.6 cm)

PAGE 70

58 5.2 M isalignment C orrection The misalignment is one of the problems that are faced when the data has been analyzed. T his issue makes the reconstructed image not clear enough to emphasize the defects, therefore, in this section the misalignment correction will be applied to one of the data that does not have defects but the eccentricity is large because of the misalignmen t. Figure 5 7 the region without defects but the shape is not circle Moreover, if the eccentricity algorithm was applied to a couple of frames, the output will not be zero as mentioned, the eccentricity will be around (0.6359) as show n in figure 5.8. Also, the 3D reconstructed image will be as figure 5.9. As a result, the correction should be applied to decrease the eccentricity to be closer to zero. This correction depends on the semi major radius, semi minor radius, and the orientati on of the ring. These parameters

PAGE 71

59 used to determine the (tform) matrix by using affine transformation. Then the image will be reshaped and squeezed to c ircle shape as shown in figure 5.10 b Figure 5 8 The eccentricity of multi frames without misalignment correction Figure 5 9 3D reconstructed image o f number of frames without apply ing misalignment correction

PAGE 72

60 Figure 5 10 a illustration of one frame that effected by misalignment and the shape is not a circle. b The same frame after applying misalignment correction and the shape is closer to circle Figure 5 11 3D reconstructed image after applying misalignment correction

PAGE 73

61 Finally, the eccentricity of the same frames after applying misalignment correction will be around (zero) which is what we expected to have a nd figure 5.1 2 illustrates the eccentricity after misalignment correction. Figure 5 12 The eccentricity after applying misalignment correction

PAGE 74

62 6. Conclusion and F uture W ork 6.1 Conclusion Optical sensor and NDE techniques are effective methods to use for studying the material discontinuities especially pipelines inner defects. Visual and optical testing are powerful to detect the deformation inside the pipes, but by using active acquisition, the result will be mo re beneficial. In chapter 2 the NDE methods have been discussed, and VT has been chosen as the primary method of this thesis. VT method depends on computer vision to detect the deformation, but the accuracy of this approach depends on the collecting data. Therefore, the optical sensor has been discussed in chapter 3. The optical sensor has two main techniques which are, passive acquisition and active acquisition. Both of methods has been explained briefly, and active acquisition has been selected as the mai n method of this thesis. Laser and structured light are classified under active acquisition as the methods that have chosen to do the experimental work. However, two different kinds of pipes have been tested to emphasize a different kind of issues. Firs t results were about a detection of different size of holes inside 3 inches pipeline, and the results can be organized as a following: Scanning with small view angle camera prevents to detect all the inner surface, and some of the important data are not captured. Scanning with large view angle camera (fisheye camera) captures all the scene, and the defects can be easily detected. One of the limitations in this type of scanning is the color dependency, which means the accuracy of the detection depends on t he color of the laser and the pipe. The other limitation is the misalignment when the prototype has some unexpected tilting because of using human hands while doing

PAGE 75

63 the scan. In chapter 5, the misalignment problem has been reduced when the eccentricity and image transformation have been done to the collected data. The structured light technique has been used as a second method to emphasize the discontinuities. In this part, a new structured light projector has been developed to ge nerate a multi colors multi rings structured light. Some lenses have been used in a certain way with a pre designed slide to build the projector. Then, a new kind of pipe has been presented to test this method; all the steps has been explained in chapter 3 as well as the results of 3D reconstructed image. The defects that have been examined is squeezed part from PE pipes. The benefit and the limitation of this method can be introduced as a following: Two different colors have been used to detect the squeeze d part, which was red and green. More colors can be easily added in order to have more colors and rings, then, the color dependency problem can be solved. The size of the pipe that can be tes ted by using this prototype is 1.85 inches or more, and by minimi zing the size of the lenses the pipe size that is needed to test can be kept to a minimum as well. The deformation can be clarified by using eccentricity calculation. The frames that have defects is having eccentricity closer to one, and the others are clo ser to zero as explained in chapter 5 when semi major and semi minor axes have been calculated and used to determine the eccentricity by using the following formula. Misalignment problem also exists in this scanning and prevents the eccentricity calcula tion from getting zero value for the areas that do not have defects. As a result, misalignment correction had been applied to minimize this problem when the semi major and semi minor axes also used to define the transformation matrix in order to squeeze th e image to be closer to circle.

PAGE 76

64 6.2 F uture W ork Optical sensors for NDE applications using structured light and 3D visualization is not a small topic to cover in one thesis, but a lot of things should be considered to develop and modify to get bett er results. Misalignment, prototype size, projection resolution, and image reconstruction need more improvement and enhancement to be more productive. For the prototype design, the number of colors and the size of the lenses can be enhanced. More than tw o colors can be added to the slide. The lenses that needed to build the prototype can be chosen smaller than 1.5 inches to scan the pipes that have a smaller diameter. Likewise, a robot can be combined to prototype to avoid the misalignment problem that oc curs because of using human hands. The robot can be designed to move by fixing steps and without tilting.

PAGE 77

65 R EFERENCES [1] CCTV Pipeline Testing Survey & Reporting with Wincan v8 Retrieved from http://www.auscamqld.com.au/ .(2013, N ov). [2] Matzkanin, D. S. (n.d.). A Brief Introduction to Nondestructive Testing. The AMMTIAC Quarterly,, 1 7 10. Retrieved from http://ammtiac.alionscience.com [3] TRAINING GUIDELINES IN NON DESTRUCTIVE TESTING TECHNIQUES. (2013). Retrieved from http://w ww.iaea.org/books [4] Liquid Penetrant and Magnetic Particle Testing at Level 2. Vienna, Austria: INTERNATIONAL ATOMIC ENERGY AGENCY. (2000). [5] https://www.tec eurolab.com/eu it/esame visivo vt.aspx [6] John Markanich, P. E. (n.d.). ELECTROMAGNETIC TESTI NG TECHNIQUES FOR QUALITY GRADING OF NONFERROUS ROD AND WIRE. Pittsburgh, PA: Foerster Instruments, Inc. [7] Introduction to Non Destructive Testing Techniques. Retrieved from http://www.eis.hu.edu.jo/ACUploads/10526/Ultrasonic%20Testing.pdf [8] Introducti on to Non Destructive Testing Techniques. (n.d.). Liquid Penetrant Testing Retrieved from http://www.eis.hu.edu.jo/ACUploads [9] Introduction to Non Destructive Testing Techniques. (n.d.). Magnetic Particle Testing Retrieved from https://eis.hu.edu.jo/up load/38000000/Magnetic%20Particle%20Testing.pdf [10] Gao L, Zai F, Su S, Wang H, Chen P, Liu L. Study and Application of Acoustic Emission Testing in Fault Diagnosis of Low Speed Heavy Duty Gears. Sensors 2011; 11(1):599 611. [11]https://www.ndee d.org/EducationResources/CommunityCollege/Other%20Methods/ AE/AE_Intro.php [12] https://www.infraredheaters.com/fundamen.html [ 13] N. Pears et al. (eds.), 3D Imaging, Analysis and Applications DOI 10.1007/978 1 4471 4063 4_2, Springer Verlag London 2012 [14] J Angelo Beraldin, F. B. Active 3D sensing. Ottawa, Ontario, Canada: Institute for Information Technology. (2000). [15] Liu, Y. D. Electromagnetic Imaging Methods for Nondestructive Evaluation Applications. 11 11774 11808. doi:10.3390/s11 1211774. (2011) [16] Curless, B. (1999). Applications of Computer Vision to Computer Graphics Retrieved from ACM SIGGRAPH: https://www.siggraph.org/publications/ newsletter/ v33n4/contributions/curless.html

PAGE 78

66 [17] Shengyong Chen, Y. F. Active Sensor Planning for Multiview Vision Tasks (1 ed.). Springer Verlag Berlin Heidelberg. doi:10.1007/978 3 540 77072 5 (2008). [18] Miquel Massot, G. O. Optical Sensors and Methods for Underwater 3D Reconstruction. pp. 31525 31557. doi:10.3390/s151229864. (2015). [19] Li, L Time of Flight Camera An Introduction. Retrieved from http://www.ti.com/3dtof. (2014). [20] Pierre Olivier, V. P. LEDDAR OPTICAL TIME OF FLIGHT SENSING TECHNOLOGY: A NEW APPROACH TO DETECTION AND RANGING. WP101 v1.3 (2015). [21] Light Detection and R anging (LiDAR). (n.d.). Portland State University. Retrieved from http://web.pdx.edu/~jduh/courses/geog493f12/Week04.pdf [22] LASER SCANNING INFRASTRUCTURE ASSETS: NEW CAPACITIES, NEW OPPORTUNITIES. Autodesk Infrastructure Solutions. Retrieved from https:/ /www.cansel.ca/images/About us/White_Papers/autodesk_whitepaper lidar%20scanning%20techniques%20and%20roi.pdf (2012). [23] Sanjeev Kumar, P. K. (IEEE). An optical triangulation method for non contact profile measurement. HEFEI UNIVERSITY OF TECHNOLOGY. do i:1 4244 0726 5/06 2006 [24] Gerig, G. Structured Lighting. Retrieved from http://www.cs.cmu.edu/afs/cs/academic/class/15385 s06/lectures/ppts/lec 17.ppt (2012). [25] 239 Retrieved from https://p gjonline.com/2012/12/27/the nature of polyethylene pipe failure/(2012, December ).