Citation
Digital image recompression analysis : Facebook

Material Information

Title:
Digital image recompression analysis : Facebook
Creator:
Pippin, Megan
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Master's ( Master of science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Music and Entertainment Industry Studies, CU Denver
Degree Disciplines:
Recording arts

Record Information

Source Institution:
University of Colorado Denver Collections
Holding Location:
Auraria Library
Rights Management:
Copyright MEGAN PIPPIN. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item has the following downloads:


Full Text
DIGITAL IMAGE RECOMPRESSION ANALYSIS: FACEBOOK
by
MEGAN PIPPIN
B.S., University of North Georgia, 2014
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirement for the degree of
Master of Science
Recording Arts
2016


This thesis for the Master of Science degree by
Megan Pippin has been approved for the Recording Arts Program by
Catalin Grigoras, Chair Jeff M. Smith Jason Lewis
December 17th, 2016


Pippin, Megan (M.S., Recording Arts)
Digital Image Recompression Analysis: Facebook Thesis directed by Assistant Professor Catalin Grigoras
ABSTRACT
Social media has become incorporated into our daily routine and has changed how we connect and communicate with other people. These sites offer slight differences in appearance and functionality, but they are primarily focused on sharing images to be viewed by friends. Whether these images are of people, places or things they tend to flood a social site and prior to uploading they can contain a large amount of embedded data. This thesis aims to identify both the effect of mobile and desktop uploading on images as well as the effect on image dimensions. These characteristics are changed by the host site, Facebook in this case, and will be analyzed into two different tests.
This thesis explores previous research into social media sites and JPEG compression while providing an overall look at Facebook. Next, the foundational steps of digital image compression will be presented. In Chapter 3, the process of data collection and preparation for this research is described. The results section follows and elaborates on the findings of both tests. Finally, future research topics are purposed for further examination.
The form and content of this abstract are approved. I recommend its publication.
in
Approved: Catalin Grigoras


DEDICATION
I would like to thank my parents for the constant push to strive for more out of life and to never accept defeat. Thank you for teaching me the value of a dollar earned and a dollar spent. You both showed me how to work for what you want out of life, ask questions and to learn from your mistakes. Thank you for continually being my biggest and loudest cheerleaders.
Justin, I want to thank you for your daily unwavering support. You gave me motivation and direction whenever it was needed. You took a great deal of my worries and stress throughout this endeavor and handled it with ease. And experiencing everyday with you by my side has been the greatest gift. I cannot wait to see where our life together takes us and you will forever be my favorite person on earth, I love you.
IV


ACKNOWLEDGEMENTS
I want to thank Jeff & Catalin for their support, suggestions and immediate responses to my every email. I have learned more than I thought possible through their incredible teaching. This field of digital forensics is such an evolving area and I am thankful for the opportunity to see it grow. I am grateful for the National Center of Media Forensics for providing a wonderful environment to learn such an exciting discipline.
Leah, you rock! You always responded to my emails with clarity and promptness. You made class deadlines and important details well-known and you were always available for my thousands of questions. You made learning at a distance that much easier, thank you.
Mr. Lewis, thank you for agreeing to be a part of my committee. I enjoyed being in of your cell phone forensics class and valued your opinion on this research. You are incredibly knowledgeable and a wonderful teacher. I hope that I can continue to learn from you throughout my career.
v


TABLE OF CONTENTS
I CHAPTER: INTRODUCTION..........................................................1
PREVIOUS RESEARCH.........................................................2
II CHAPTER: OVERVIEW.............................................................4
EASE OF EDITING...........................................................7
THE SCIENTIFIC WORKING GROUPS.............................................8
JPEG OVERVIEW.............................................................9
III CHAPTER: PREP STAGE.........................................................12
FIRST TEST SETUP.........................................................16
SECOND TEST SETUP........................................................17
IV CHAPTER: RESULTS.............................................................18
TEST ONE.................................................................18
TEST TWO: DESKTOP RESIZED................................................27
TEST TWO: MOBILE RESIZED.................................................29
ADDITONAL OBSERVATIONS...................................................29
V CHAPTER: DISCUSSION..........................................................33
FUTURE RESEARCH..........................................................34
REFERENCES......................................................................35
VI


LIST OF TABLES
1.1 Exiftool Output from Original Image...........................................3
3.1 Filename, Image Dimension, Quantization Tables and Exiftool Metadata Dump for
Original Image 1 (FB Recompression Test)..........................................15
4.1 Results Matrix for Mobile and Desktop Processing.............................17
4.2 Mobile Downloaded Schema; Facebook Filename, Image Dimension, Exiftool Metadata
Dump and Quantization Tables..................................................18
4.3 Mobile Upload, Desktop Download Schema; Facebook Filename, Image Dimension,
Exiftool Metadata Dump and Quantization Tables................................21
4.4 Desktop Upload, Desktop Download Schema; Facebook Filename, Image Dimension,
Exiftool Metadata Dump and Quantization Tables................................24
4.5 Resized Images Spreadsheet with Aspect Ratios................................27
4.6 Aspect Ratios for Mobile Uploaded Resized Image Sizes........................30
4.7 Facebook Filename with Ending Variables and Image Dimensions.................31
vii


LIST OF FIGURES
2.1 Steps in Digital Image Compression; RGB through Coding.......................11
3.1 Test One Data Matrix.........................................................16
viii


LIST OF ABBREVIATIONS
CFA Color Filter Array
DCT- Discrete Cosine Transform
DSLR Digital Single-Lens Reflex
ELA Error Level Analysis
GPS Global Positioning System
JPEG Joint Photographic Experts Group
RLC Run Length Coding
SWGDE Scientific Working Group on Digital Evidence SWGIT Scientific Working Group on Imaging Technology
QMS Quality Management System


CHAPTER I
INTRODUCTION
Facebook is one of the most well-known social media websites with an estimated "1.13 billion daily active users on average for June 2016" [4], At its inception on February 4, 2004 Facebook began with the intent of joining Harvard college students [1], Facebook now can connect billions of people in seconds to not only acquaintances and friends, but also to businesses, celebrities and political figures. Facebook has evolved in several ways and provides additional features to users but regardless of its growth it has continually allowed for the uploading and downloading of images. These images can be viewed on several devices such as desktops, laptops, cellphones and tablets. Facebook can be accessed through these devices via a downloadable application or directly through a web browser.
These photos are often uploaded, shared and then collected in the user's profile for later viewing. Images that have been uploaded and made viewable provide useful information to aid criminals in inferring the location of your home, family information and whether you may be home or on vacation, leaving your possessions unguarded. It is possible to protect certain elements of your page through privacy settings but these are often overlooked [7],
Facebook users can encounter phony profiles attempting to pose as friends or family members to gain your trust and swindle you out of your money or valuable information. Although Facebook states to have around one billion daily users, it is estimated that up to
8.7% of those could be fake accounts [6], It is important to decipher profiles of legitimate


people and those of deceptive individuals, this can be accomplished by using digital forensics.
Forensic digital image analysis can aid in locating and interpreting the embedded data encoded inside of digital images, the Scientific Working Group on Digital Evidence (SWGDE) defines multimedia evidence as "Analog or digital media, including but not limited to, film, tape, magnetic and optical media, and/or the information contained therein" [13].
PREVIOUS RESEARCH
There has been previous research investigating the forensic analysis of digital image compression and the impact social media applications has on processed images. Wu et al. proposed a solution to "compress a set of JPEG coded correlated images without loss" by removing inter and intra repetition. This is done by decoding an image collection and a prediction structure is compiled that can determine any commonality across the images. Those similarities can be exploited for the collection rather than each individual picture.
This type of analysis can be beneficial and reduce cost for storage of large social media sites [15].
In regards to different social networking sites, Castiglione et al. analyzed at three different social sites in 2011. They collected various sets of images that included different image resolutions and sizes as well as thumbnails and profile pictures. They determined that compression removed EXIF metadata from processed image and different quality images allow limited pixel resolution [3].
2


Marrion explored the discrete cosine transform (DCT) map and error level analysis (ELA)
of images processed through Facebook to detect any indications of manipulation. The results concluded that once a tampered image was compressed by Facebook the previous manipulation was usually undetectable. The compression affected the surrounding neighboring pixels and masked the alterations [8],
3


CHAPTER II
OVERVIEW
The embedded data typically in an original image can be found in the following Table. This is a partial output generated by using Exiftool. The table shows the file size, file type, make and model of the camera, time stamp and the global positioning system (GPS) information.
Table 1.1: Exiftool Output from Original Image
ExifToolVersion = 10.25 FileName = 1920xl440i.jpg
Directory = C:/Users/megan_000/Desktop/Resized Images FileSize = 688428
FileModifyDate = 1476455070.39326 FileAccessDate = 1476455070.38826 FileCreateDate = 1476455070.38826 FilePermissions = 33206 FileType = JPEG FileTypeExtension = JPG MIMEType = image/jpeg JPEG APP0 (14 bytes):
+ [BinaryData directory, 9 bytes]
| JFIFVersion = 11 | Resolutionllnit = 1 | XResolution = 72 | YResolution = 72 JPEG APP1 (12284 bytes):
ExifByteOrder = MM + [IFD0 directory with 11 entries]
| 0) Make = Apple | 1) Model = iPhone 6s | 2) Orientation = 6 | 3) XResolution = 72 (72/1)
| 4) YResolution = 72 (72/1)
| 5) Resolutionllnit = 2 | 6) Software = 9.3.4
4


Table 1.1: Exiftool Output from Original Image, Cont'd
| 7) ModifyDate = 2016:09:06 06:46:07 | 8) YCbCrPositioning = 1 | 9) ExifOffset (SubDirectory) >
| + [ExifIFD directory with 32 entries]
| | 0) ExposureTime = 0.06666666667 (1/15)
| | 1) FNumber = 2.2 (11/5)
| | 2) ExposureProgram = 2
| | 3) ISO = 500
| | 4) ExifVersion = 0221
| | 5) DateTimeOriginal = 2016:09:06 06:46:07
| | 6) CreateDate =2016:09:06 06:46:07
| | 7) ComponentsConfiguration = 12 3 0
| | 8) ShutterSpeedValue = 3.906969876 (13229/3386)
| | 9) ApertureValue = 2.275007125 (7983/3509)
| | 10) BrightnessValue =-1.80421074 (-17739/9832)
| | 11) ExposureCompensation = 0 (0/1)
| | 12) MeteringMode = 5 | | 13) Flash = 16
| | 14) FocalLength = 4.15 (83/20)
| | 15) SubjectArea = 2015 1511 2217 1330 | | 16) MakerNoteApple (SubDirectory) >
| | + [MakerNotes directory with 11 entries]
| | | 0) Apple_0x0001 = 4
| | | 1) Apple_0x0002 = bplistOOO....X]]]XRQRYAkqib..]_cbZTVUAfx{xq..A]cd]XW]Ak... |.[snip] | | | 2) RunTime (SubDirectory)-->
||| + [Binary PLIST directory]
| | | | RunTimeFlags = 1 | | | | RunTimeValue = 490495960531250 | | | | RunTimeEpoch = 0 | | | | RunTimeScale = 1000000000 | | | 3) Apple_0x0004 = 1 | | | 4) Apple_0x0005 = 80 | | | 5) Apple_0x0006 = 92 | | | 6) Apple_0x0007 = 1
| | | 7) Apple_0x0008 = 0.03742304133 -0.7731738573 -0.603210609 (930/24851-1810/23 [snip]
| | | 8) Apple_0x0009 = 4371 | | | 9) Apple OxOOOe = 0 | | | 10) Apple_0x0014 = 4 | | 17) SubSecTimeOriginal = 132 | | 18) SubSecTimeDigitized = 132


Table 1.1: Exiftool Output from Original Image, Cont'd
| | 19) FlashpixVersion = 0100
| | 20) ColorSpace = 1
| | 21) ExiflmageWidth = 4032
| | 22) ExiflmageHeight = 3024
| | 23) SensingMethod = 2
| | 24) SceneType = 1
| | 25) ExposureMode = 0
| | 26) WhiteBalance = 0
| | 27) FocalLengthln35mmFormat = 29
| | 28) SceneCaptureType = 0
| | 29) Lenslnfo = 4.15 4.15 2.2 2.2 (83/20 83/20 11/5 11/5) | | 30) LensMake = Apple
| | 31) LensModel = iPhone 6s back camera 4.15mm f/2.2 | 10) GPSInfo (SubDirectory) >
| + [GPS directory with 15 entries]
| | 0) GPSLatitudeRef = N
| | 1) GPSLatitude = 39 39 54.25 (39/1 39/1 5425/100)
| | 2) GPSLongitudeRef = W
| | 3) GPSLongitude = 104 56 58.78 (104/1 56/1 5878/100)
| | 4) GPSAItitudeRef = 0 | | 5) GPSAItitude = 1630.50495 (164681/101)
| | 6) GPSTimeStamp = 12 46 6 (12/1 46/1 600/100)
| | 7) GPSSpeedRef = K | | 8) GPSSpeed = 0 (0/1)
| | 9) GPSlmgDirectionRef = T | | 10) GPSlmgDirection = 322.2258065 (9989/31)
| | 11) GPSDestBearingRef = T | | 12) GPSDestBearing = 322.2258065 (9989/31)
| | 13) GPSDateStamp = 2016:09:06 | | 14) GPSHPositioningError= 10 (10/1)
+ [IFD1 directory with 6 entries]
| 0) Compression = 6 | 1) XResolution = 72 (72/1)
| 2) YResolution = 72 (72/1)
| 3) Resolutionllnit = 2 | 4) ThumbnailOffset = 2014 | 5) ThumbnailLength = 9017 JPEG DQT (65 bytes):
JPEG DQT (65 bytes):
JPEG SOFO (15 bytes):
ImageWidth = 1920


ImageHeight = 1440 EncodingProcess = 0 BitsPerSample = 8 ColorComponents = 3 JPEG DHT (28 bytes):
JPEG DHT (72 bytes):
JPEG DHT (26 bytes):
JPEG DHT (47 bytes):
JPEG SOS____________________________________________________________________________
Table 1.1: Exiftool Output from Original Image, Cont'd
This information can be useful in an investigation and lend additional data that can be analyzed for authenticity purposes. This thesis presents research into how this material is changed when images are recompressed by Facebook.
EASE OF EDITING
Criminals currently do not run at a disadvantage when it comes to creating false images. Cellphones are cheaper than expensive DSLR (Digital Single-Lens Reflex) cameras and still produce a high-resolution image. Computers, laptops and rapid processing tablets are accessible in stores and online. Digital images are readily available on the internet for free download or for purchase through services such as Adobe Stock. The only difference is the motivation and intention of the person who produces the altered images. Unfortunately, manipulated images have affected well-known people such as presidents, politicians and celebrities. The main goal of these pictures is to skew the overall meaning or to achieve some type of political gain. The ability to change a digital image is not a new phenomenon but it does require editing software and the tenacity to learn this time-consuming skill. There are several programs for purchase such as Adobe Photoshop and Corel Paintshop that are very comprehensive and offer advanced features for manipulation
7


but programs can also be downloaded from the internet that can accomplish similar results.
These software options paired with vast amounts of "how-to" videos on YouTube and informational forums and creating a manipulated image can be accomplished by someone with no formal training.
THE SCIENTIFIC WORKING GROUPS
As mentioned before what classifies as multimedia evidence and how its examination should be carried out are crucial to producing a clear analysis. One source for best practices in this discipline the Scientific Working Group on Digital Evidence (SWGDE) which is formed of committees which produce documents to guide examinations on various digital and multimedia devices, storage units, and files. The group "brings together organizations actively engaged in the field of digital and multimedia evidence to foster communication and cooperation as well as to ensure quality and consistency within the forensic community" [13]. Ultimately, creating a learning environment to produce the best techniques and adapting to new technology as it pertains to an investigation. Although SWGDE does not provide accreditation to labs it does offer documents to be incorporated into labs Quality Management Systems (QMS) [14].
There is another committee devoted to preparing documents to aid examiners in their investigations of digital media. The Scientific Working Group on Imaging Technology (SWGIT) has created a beneficial document regarding forensic image analysis. These documents focus on proper procedure for seizing, processing and analyzing forensic digital
8


imagery. Although now defunct, SWGIT working groups have been formed and are active
within SWDGE.
JPEG OVERVIEW
In 1992, the Joint Photographic Experts Group, (JPEG) committee gathered to study color imagery and develop the well-known standard for image compression [10]. JPEG compression is lossy which SWGDE defines as "compression in which data is lost and cannot be retrieved in its original form" [13]. JPEG achieves the best results on areas of images that contain low frequency changes instead of stark color differences. For this research, the concern is with lossy compression and its effect on digital images processed through Facebook.
It is important to mention that there are many degrees of JPEG compression that can be applied to images that result in a range of overall quality degradation. The amount of quantization error is based on the selection of quantization tables which can roughly translate into a quality level that ranging from 1 (the lowest quality) to 99, (highest quality).
Typically, the amount of pixel information that is destroyed is not discernable to the human eye unless you move lower in quality factors. However, once a file is saved and the compression algorithm is applied continuously over several iterations the amount of visual degradation will become more visible. Following is a general overview of the JPEG compression process.
Color is received from the color filter array (CFA) commonly as RGB (red, blue, green) color values and then converted into YCbCr for compression. "Y" represents the
9


(luma) or luminance value which is simply the intensity of brightness in an image. "Cb" and
"Cr" represent the two chrominance values of blue and red. This color space is more easily exploited for compression and can be stored with a higher image resolution and treated as separate entities for greater efficiency [11],
Figure 2.1 depicts the process that follows YCbCr conversion. The boxes show progression into an image being split into 8x8 blocks of pixels. Those areas are then subjected to a Discrete Cosine Transform (DCT) which "converts pixel values into transform coefficients corresponding to spatial frequencies" [5].
After DCT is applied, the values at each pixel are quantized; the resulting value can be correlated to the weight each pixel provides in producing details of the overall image. This step compresses the image by removing the less significant, less noticeable changes in values between pixels and the more pixels that can be made uniform, the smaller the resulting file. This results in removing high frequency details; a change which is less noticeable to the human eye. Information removed during the quantization phase cannot be retrieved during decompression and contribute most significantly to JPEG compression's lossy [2].
The pixels now need to be rearranged from an 8x8 block into a one-dimensional array for subsequent Run Length Coding (RLE) and Huffman coding. This is achieved by collecting the values starting at the upper left corner of the block and creating a zig-zag pattern all through the block until the final piece is obtained from the bottom right corner.
10


Figure 2.1: Steps in Digital Image Compression; RGB through Coding
11


CHAPTER III
PREPARATION STAGE
In this research, transcoding methods applied to images that are processed through Facebook were analyzed in two different tests utilizing both desktop and mobile devices. This transcoding step involves recompression of the image and the creation of new metadata in the resulting JPEG file. Both tests used the same desktop, ASUS laptop and a Samsung Galaxy Note 5 cellphone. A test Facebook profile and Google email, Gmail, account was created to separate images and emails from personal accounts. The data collected for analysis included the quantization tables, image size (height and width dimensions) and metadata structure; both before uploading to Facebook and after.
Both tests utilized programs that analyze embedded data and display it in a readable format. These programs include Medialnfo which was used to determine image size on processed images from Facebook, it also "is a convenient unified display of the most relevant technical and tag data for video and audio files" [9], A script was created in MATLAB to extract the quantization tables and metadata structure for each image. Finally, Exiftool was used to view metadata structure and individual elements.
The following Table 3.1 shows the filename, image dimension, quantization tables and the Exiftool metadata dump of an original file prior to Facebook upload. This image was used in the first test for Facebook recompression analysis.
12


Table 3.1: Filename, Image Dimension, Quantization Tables and Exiftool Metadata
Dump for Original Image 1 (FB Recompression Test)
Facebook Filename: 20160906_121154.jpg
Image Dimension: 2988 x 5312
ExifToolVersion = 10.25 FileName = 20160906_121154.jpg
Directory = C:/Users/megan_000/Desktop/Original Image 1 FileSize = 5015528
FileModifyDate = 1473178697.94831 FileAccessDate = 1480562476.34245 FileCreateDate = 1480562476.34245 FilePermissions = 33206 FileType = JPEG FileTypeExtension =JPG MIMEType = image/jpeg
JPEG APP1 (15990 bytes):
ExifByteOrder = II + [IFD0 directory with 13 entries]
| 0) ImageWidth = 5312
| 1) ImageFleight = 2988
| 2) Make = Samsung
| 3) Model = SAMSUNG-SM-N920A
| 4) Orientation = 6
| 5) XResolution = 72 (72/1)
| 6) YResolution = 72 (72/1)
| 7) Resolutionllnit = 2
| 8) Software = N920AUCS3BPH4
| 9) ModifyDate = 2016:09:06 12:11:54
| 10) YCbCrPositioning = 1
| 11) ExifOffset (SubDirectory) >
| + [ExifIFD directory with 26 entries]
| | 0) ExposureTime = 0.1 (1/10)
| | 1) FNumber= 1.9 (19/10)
| | 2) ExposureProgram = 2 | | 3) ISO = 250 | | 4) ExifVersion = 0220 | | 5) DateTimeOriginal =2016:09:06 12:11:54 | | 6) CreateDate = 2016:09:06 12:11:54 | | 7) ShutterSpeedValue = 3.32 (332/100)
13


Table 3.1: Filename, Image Dimension, Quantization Tables and Exiftool Metadata Dump for Original Image 1 (FB Recompression Test), Cont'd
| | 8) ApertureValue = 1.85 (185/100)
| | 9) BrightnessValue = -1.06 (-106/100)
| | 10) ExposureCompensation = 0 (0/10)
| | 11) MaxApertureValue = 1.85 (185/100)
| | 12) MeteringMode = 2 | | 13) Flash =0
| | 14) FocalLength = 4.3 (430/100)
| | 15) MakerNoteUnknown (SubDirectory) >
| | + [MakerNotes directory with 7 entries]
| | | 0) Unknown OxOOOl = 0100 | | | 1) Unknown_0x0002 = 73728 | | | 2) UnknownOxOOOc = 0 | | | 3) UnknownOxOOlO = undef (0/0)
| | | 4) Unknown_0x0040 = 0 | | | 5) Unknown_0x0050 = 1 | | | 6) UnknownOxOlOO = 0 | | 16) UserComment =
| | 17) FlashpixVersion = 0100 | | 18) ColorSpace = 1 | | 19) ExiflmageWidth = 5312 | | 20) ExiflmageHeight = 2988 | | 21) ExposureMode = 0 | | 22) WhiteBalance = 0 | | 23) FocalLengthln35mmFormat = 28 | | 24) SceneCaptureType = 0
| | 25) ImageUniquelD = B16LSIA00SM B16LSIL03SM.
| 12) GPSInfo (SubDirectory) >
| + [GPS directory with 9 entries]
| | 0) GPSVersionID = 2 2 00
| | 1) GPSLatitudeRef = N
| | 2) GPSLatitude = 34 6 31 (34/1 6/1 31/1)
| | 3) GPSLongitudeRef = W | | 4) GPSLongitude = 83 52 33 (83/1 52/1 33/1)
| | 5) GPSAItitudeRef = 1 | | 6) GPSAItitude = 0 (0/1)
| | 7) GPSTimeStamp = 16 11 38 (16/1 11/1 38/1)
| | 8) GPSDateStamp = 2016:09:06 + [IFD1 directory with 9 entries]
| 0) ImageWidth = 512
14


Table 3.1: Filename, Image Dimension, Quantization Tables and Exiftool Metadata Dump for Original Image 1 (FB Recompression Test), Cont'd
| 1) ImageHeight = 288 | 2) Compression = 6 | 3) Orientation = 6 | 4) XResolution = 72 (72/1)
| 5) YResolution = 72 (72/1)
| 6) Resolutionllnit = 2 | 7) ThumbnailOffset = 1154 | 8) ThumbnailLength = 14830 JPEG APP4 (13142 bytes):
JPEG APP5 (24 bytes):
JPEG DHT (416 bytes):
JPEG DQT (130 bytes):
JPEG SOFO (15 bytes):
ImageWidth = 5312 ImageHeight = 2988 EncodingProcess = 0 BitsPerSample = 8 ColorComponents = 3 JPEG SOS
Samsung trailer (67 bytes at offset 0x4c87a5): Samsung_Trailer_0x0a01-name = ImageUTCData TimeStamp = 1473178315139_________________________
3 4 4 4 4 3 3 4
4 4 4 4 3 3 3 4
4 4 4 4 3 3 4 5
4 3 3 3 3 4 5 6
3 3 3 3 5 6 7 8
3 3 3 4 6 8 9 12
3 3 4 5 7 9 12 15
4 4 5 6 8 12 15 21


3 3 3 3 5 6 7 8
3 3 3 4 6 8 9 12
3 3 4 5 7 9 12 15
4 4 5 6 8 12 15 21
Table 3.1: Filename, Image Dimension, Quantization Tables and Exiftool Metadata
Dump for Original Image 1 (FB Recompression Test), Cont'd
FIRST TEST SETUP
Currently, there are two different ways to upload an image through Facebook, via a desktop with an internet browser and a mobile device either through a browser or with the Facebook application. With these options, an image can be uploaded as a single image or in a collection of images, called an album. Downloading capabilities are slightly more limited; images can be downloaded using both a mobile and desktop but at the time of the research the ability to download an album through a mobile device is not available. Since these options allow for twelve different scenarios a matrix was created to input all data and properly explore all options seen that be seen in Figure 3.1 below.
16


Method
mobile download (1 al a lime] Desktop download (1 at a time} Desktop download (all at once)
mobile upload (1 at a time)
mobile upload (dll dt once}
Desktop upload ji at a time)
Desktop u pload (all at once}
Figure 3.1: Test One Data Matrix
While considering the avenues above, five original images in the same portrait orientation were obtained from the cellphone to be used in the first test. The embedded data from each photo was retrieved for comparison against the processed images. Main folders were created to represent each image and each permutation of mobile and desktop uploading and downloading were given their own subfolders. This was used to keep all obtained data separate for analysis that will later be explained in Chapter 4.
SECOND TEST SETUP
The second test was composed of taking a single image that originated from an Apple iPhone 6S and rescaling it into 56 common image dimensions. These sizes ranged from the smallest, 640 x 480, to the largest, 5120 x 3200. Each image was uploaded, both
17


from a desktop and mobile device, and downloaded to a desktop computer to observe the
pattern Facebook uses to resize uploaded images.
18


CHAPTER IV
RESULTS
TEST ONE
After analyzing the data from the first test it was determined that there is no difference in changes to an image between uploading it individually or in an album. The main deviations occur through three main schemas present throughout the five images. The first pattern is that mobile downloaded images, regardless of how they are uploaded, are recompressed with specific quantization tables using Baseline encoding and produce a distinct but repeated metadata structure composed of 65 elements. The other two schema relate to images downloaded to a desktop computer. The first of these, the second overall schema, was seen in images that were uploaded from a mobile device then downloaded to a desktop computer; these are recompressed with distinct quantization tables using Progressive encoding and a metadata structure of 67 elements. Finally, images were uploaded and downloaded from a desktop computer are Progressive encoded as well but have variable quantization tables roughly equivalent to a JPEG quality factor of 87 along with a repeated metadata structure of 81 elements. A summary of these can be found in Table 4.1 below. The size, quantization tables, and metadata dump for an investigated image (the same presented in Chapter 3, page 12 above) in each of these three scenarios follows.
19


Results mobile download (1 at a time} Desktop download (1 at a time) Desktop download (all at once)
mobile upload (1 at a time) Baseline DCT JPEG Quality: 93 Progressive OCT JPEG Quality: 64 Progressive OCT JPEG Quality:64
mobile upload (all at once) Baseline OCT JPEG Quality:93 Progressiva OCT JPEG Quality:64 Progressive OCT JPEG Quality: 64
Desktop upload (1 at a time) l Baseline DCT JPEG Quality: 93 Progressive DCT JPEG Quality: 87* Progressive DCT JPEG Quality: 87*
Desktop upload (all at once) Baseline DCT JPEG Quality. 93 Progressive OCT JPEG Quality;87* Progressive DCT JPEG Quality: 87*
VttJuaMLri' v: £2 Metadata furies -
* Variable for different images but same for one image
Table 4.1: Results Matrix for Mobile and Desktop Processing
In each of the following tables all three schemas and their respective filename, image dimension, quantization tables and Exiftool Metadata dump. The first table shows the mobile downloaded image, the second depicts the desktop downloaded output and the final image will display the desktop downloaded pattern.
Table 4.2: Mobile Downloaded Schema; Facebook Filename, Image Dimension, Exiftool
Metadata Dump and Quantization Tables
Facebook Filename: 14238300_137157913404958_5593732394586807444_n.jpg
Image Dimension: 528 x 960____________________________________________
ExifToolVersion = 10.25
FileName = 14238300_137157913404958_5593732394586807444_n.jpg Directory = C:/Users/megan_000/Desktop/Mobiie Upload, Mobile Download FileSize = 147946
FileModifyDate = 1473541535.57959 FileAccessDate = 1480564817.46776 FileCreateDate = 1480564817.46776 FilePermissions = 33206 FileType = JPEG FileTypeExtension = JPG
20


Table 4.2: Mobile Downloaded Schema; Facebook Filename, Image
Dimension, Exiftool Metadata Dump and Quantization Tables, Cont'd
MIMEType = image/jpeg JPEG APP2 (538 bytes):
ICC Profile chunk 1 of 1
+ [ICC Profile directory with 10 entries, 524 bytes]
| ProfileHeader (SubDirectory) >
| + [BinaryData directory, 128 bytes]
| | ProfileCMMType = Icms
| | ProfileVersion = 528
| | ProfileClass = mntr
| | ColorSpaceData = RGB
| | ProfileConnectionSpace = XYZ
| | ProfileDateTime = 2012 1 25 3 41 57
| | ProfileFileSignature = acsp
| | PrimaryPlatform = APPL
| | CMMFIags = 0
| | DeviceManufacturer =
| | DeviceModel =
| | DeviceAttributes = 0 0 | | Renderinglntent = 0
| | ConnectionSpacellluminant = 0.9642 10.82491 | | ProfileCreator = Icms | | ProfilelD = 0000000000000000
0) ProfileDescription = c2
1) ProfileCopyright = FB
2) MediaWhitePoint = 0.9642 10.82491
3) MediaBlackPoint = 0.01205 0.0125 0.01031
4) RedMatrixColumn = 0.43607 0.22249 0.01392
5) GreenMatrixColumn = 0.38515 0.71687 0.09708
6) BlueMatrixColumn = 0.14307 0.06061 0.7141
7) RedTRC = curv....c...k...?.Q.4!.).2.;.F.Qw].kpz.... | ,i.}...0..
8) GreenTRC = curv...c...k...?.Q.4!.).2.;.F.Qw].kpz.... | ,i.}...0.
9) BlueTRC = curv...c...k...?.Q.4!.).2.;.F.Qw].kpz.... | ,i.}...0..
JPEG APP0 (14 bytes):
+ [BinaryData directory, 9 bytes]
JFIFVersion = 11 Resolutionllnit = 0 XResolution = 1 YResolution = 1
JPEG APP13 (52 bytes):
+ [Photoshop directory, 38 bytes]


Table 4.2: Mobile Downloaded Schema; Facebook Filename, Image
Dimension, Exiftool Metadata Dump and Quantization Tables, Cont'd
| IPTCData (SubDirectory) >
| + [IPTC directory, 25 bytes]
| | CurrentIPTCDigest = ....B..V.F.W....
| | -- IPTCApplication record
| | OriginalTransmissionReference = dr8ylh8EV8yvmlHXOYQb JPEG DQT (65 bytes):
JPEG DQT (65 bytes):
JPEG SOF2 (15 bytes):
ImageWidth = 528 ImageHeight = 960 EncodingProcess = 2 BitsPerSample = 8 ColorComponents = 3 JPEG DHT (26 bytes):
JPEG DHT (25 bytes):
JPEG SOS
3 2 3 3 4 5 10 14
2 2 3 3 4 7 13 18
2 3 3 4 7 11 16 19
3 4 5 6 11 13 17 20
5 5 8 10 14 16 21 22
8 12 11 17 22 21 24 20
10 12 14 16 21 23 24 21
12 11 11 12 15 18 20 20
3 4 5 9 20 20 20 20
4 4 5 13 20 20 20 20
5 5 11 20 20 20 20 20
9 13 20 20 20 20 20 20
20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20
Table 4.2: Mobile Downloaded Schema; Facebook Filename, Image Dimension, Exiftool Metadata Dump and Quantization Tables, Cont'd
22


Table 4.3: Mobile Upload, Desktop Download Schema; Facebook Filename, Image
Dimension, Exiftool Metadata Dump and Quantization Tables
Facebook Filename: 14199208_132277663892983_6048687515817584365_n(l)
Image Dimension: 528 x 960_______________________________________________
ExifToolVersion = 10.25
FileName = 14199208_132277663892983_6048687515817584365_n(l).jpg Directory = C:/Users/megan_000/Desktop/Mobile Upload, Desktop Download FileSize = 147946
FileModifyDate = 1473352812.10288 FileAccessDate = 1480563970.24043 FileCreateDate = 1480563970.24043 FilePermissions = 33206 FileType = JPEG FileTypeExtension = JPG MIMEType = image/jpeg JPEG APP2 (538 bytes):
ICC Profile chunk 1 of 1
+ [ICC Profile directory with 10 entries, 524 bytes]
| ProfileFleader (SubDirectory) >
| + [BinaryData directory, 128 bytes]
| | ProfileCMMType = Icms
| | ProfileVersion = 528
| | ProfileClass = mntr
| | ColorSpaceData = RGB
| | ProfileConnectionSpace = XYZ
| | ProfileDateTime = 2012 1 25 3 41 57
| | ProfileFileSignature = acsp
| | PrimaryPlatform = APPL
| | CMMFIags = 0
| | DeviceManufacturer =
| | DeviceModel =
| | DeviceAttributes = 0 0 | | Renderinglntent = 0
| | ConnectionSpacellluminant = 0.9642 10.82491
| | ProfileCreator = Icms
| | ProfilelD = 0000000000000000
| 0) ProfileDescription = c2
| 1) ProfileCopyright = FB
| 2) MediaWhitePoint = 0.9642 10.82491
23


Table 4.3: Mobile Upload, Desktop Download Schema; Facebook
Filename, Image Dimension, Exiftool Metadata Dump and Quantization
Tables, Cont'd
| 3) MediaBlackPoint = 0.01205 0.0125 0.01031 | 4) RedMatrixColumn =0.43607 0.22249 0.01392 | 5) GreenMatrixColumn =0.38515 0.71687 0.09708 | 6) BlueMatrixColumn = 0.14307 0.06061 0.7141
| 7) RedTRC = curv.....c...k...?.Q.4!.).2.;.F.Qw].kpz....|.i.}...0..
| 8) GreenTRC = curv.....c...k...?.Q.4!.).2.;.F.Qw].kpz.... | ,i.}...0..
| 9) BlueTRC = curv....c...k...?.Q.4!.).2.;.F.Qw].kpz....|.i.}...0..
JPEG APP0 (14 bytes):
+ [BinaryData directory, 9 bytes]
| JFIFVersion = 11 | ResolutionUnit = 0 | XResolution = 1 | YResolution = 1 JPEG APP13 (52 bytes):
+ [Photoshop directory, 38 bytes]
| IPTCData (SubDirectory) >
| + [IPTC directory, 25 bytes]
| | CurrentIPTCDigest = E:8AAz..x.a..3.
| | -- IPTCApplication record
| | OriginalTransmissionReference = uADamwBoBFIvuzcl85-ha JPEG DQT (65 bytes):
JPEG DQT (65 bytes):
JPEG SOF2 (15 bytes):
ImageWidth = 528 ImageFleight = 960 EncodingProcess = 2 BitsPerSample = 8 ColorComponents = 3 JPEG DHT(26 bytes):
JPEG DHT(25 bytes):
JPEG SOS


Table 4.3: Mobile Upload, Desktop Download Schema; Facebook Filename, Image Dimension, Exiftool Metadata Dump and Quantization Tables, Cont'd
Table 4.4: Desktop Upload, Desktop Download Schema; Facebook Filename, Image Dimension, Exiftool Metadata Dump and Quantization Tables
Facebook Filename: 14249740_137075313413218_424651193658018534_o Image Dimension: 1152 x 2048 ExifToolVersion = 10.25
FileName = 14249740_137075313413218_424651193658018534_o.jpg Directory = C:/Users/megan_000/Desktop/Desktop Upload, Desktop Download FileSize = 363398
FileModifyDate = 1473538212.92549 FileAccessDate = 1480564002.64724 FileCreateDate = 1480564002.64724 FilePermissions = 33206 FileType = JPEG FileTypeExtension = JPG MIMEType = image/jpeg JPEG APP0 (14 bytes):
25


Table 4.4: Desktop Upload, Desktop Download Schema; Facebook Filename,
Image Dimension, Exiftool Metadata Dump and Quantization Tables, Cont'd
+ [BinaryData directory, 9 bytes]
| JFIFVersion = 12 | ResolutionUnit = 0 | XResolution = 1 | YResolution = 1 JPEG APP13 (154 bytes):
+ [Photoshop directory, 140 bytes]
| IPTCData (SubDirectory) >
| + [IPTC directory, 128 bytes]
| | CurrentIPTCDigest = -R/f.?:...#F.K..
| | -- IPTCApplication record
| | OriginalTransmissionReference = O40Kf62SC2BFolgP3P00 | | Speciallnstructions =
FBMD01000a9e0d0000b28200009e2e0100b63b0100ac4b010075df01000[snip]
JPEG APP2 (3062 bytes):
ICC Profile chunk 1 of 1
+ [ICC Profile directory with 16 entries, 3048 bytes]
| ProfileFleader (SubDirectory) >
| + [BinaryData directory, 128 bytes]
| | ProfileCMMType =
| | ProfileVersion = 512
| | ProfileClass = mntr
| | ColorSpaceData = RGB
| | ProfileConnectionSpace = XYZ
| | ProfileDateTime = 2009 3 27 21 36 31
| | ProfileFileSignature = acsp
| | PrimaryPlatform =
| | CMMFIags = 0 | | DeviceManufacturer =
| | DeviceModel =
| | DeviceAttributes = 10 | | Renderinglntent = 0
| | ConnectionSpacellluminant = 0.9642 10.82491 | | ProfileCreator =
| | ProfilelD = 41 248 61 222 175 242 85 174 120 66 250 228 202 131 57 13 | 0) ProfileDescription = sRGB IEC61966-2-1 black scaled | 1) BlueMatrixColumn = 0.14307 0.06061 0.7141
| 2) BlueTRC = curv..#(-27;@EJOTYAchmrw|........................%[snip]
| 3) DeviceModelDesc = IEC 61966-2-1 Default RGB Colour Space Srgb | 4) GreenMatrixColumn =0.38515 0.71687 0.09708
26


Table 4.4: Desktop Upload, Desktop Download Schema; Facebook Filename,
Image Dimension, Exiftool Metadata Dump and Quantization Tables, Cont'd
| 5) GreenTRC = curv.#(-27;@EJOTYAchmrw|.......................[snip]
| 6) Luminance = 0 80 0 | 7) Measurement (SubDirectory) -->
| + [BinaryData directory, 36 bytes]
| | MeasurementObserver = 1
| | MeasurementBacking = 0 0 0
| | MeasurementGeometry = 0
| | MeasurementFlare = 0
| | Measurementllluminant = 2
| 8) MediaBlackPoint = 0.01205 0.0125 0.01031
| 9) RedMatrixColumn =0.43607 0.22249 0.01392
| 10) RedTRC = curv..#(-27;@EJOTYAchmrw|.......................%.[snip]
| 11) Technology = CRT
| 12) ViewingCondDesc = Reference Viewing Condition in IEC 61966-2-1 | 13) MediaWhitePoint = 0.9642 1 0.82491
| 14) ProfileCopyright = Copyright International Color Consortium, 2009 | 15) ChromaticAdaptation = 1.04791 0.02293 -0.0502 0.0296 0.99046 -0.01707 -0.00925 [snip]
JPEG DQT (65 bytes):
JPEG DQT (65 bytes):
JPEG SOF2 (15 bytes):
ImageWidth = 1152 ImageHeight = 2048 EncodingProcess = 2 BitsPerSample = 8 ColorComponents = 3 JPEG DHT (25 bytes):
JPEG DHT (24 bytes):
JPEG DHT (24 bytes):
JPEG SOS
6 4 4 6 10 16 20 24
5 5 6 8 10 23 24 22
6 5 6 10 16 23 28 22
6 7 9 12 20 35 32 25
7 9 15 22 27 44 41 31
10 14 22 26 32 42 45 37
20 26 31 35 41 48 48 40
29 37 38 39 45 40 41 40


7 7 10 19 40 40 40 40
7 8 10 26 40 40 40 40
10 10 22 40 40 40 40 40
19 26 40 40 40 40 40 40
40 40 40 40 40 40 40 40
40 40 40 40 40 40 40 40
40 40 40 40 40 40 40 40
40 40 40 40 40 40 40 40
Table 4.4: Desktop Upload, Desktop Download Schema; Facebook Filename, Image Dimension, Exiftool Metadata Dump and Quantization Tables, Cont'd
TEST TWO RESULTS: DESKTOP RESIZED
After examining the images resized by Facebook, it was determined in the particular test that in these scenarios with these devices, the dimension cutoff for mobile picture width is 1440 pixels and desktop picture width is 2048 pixels. The following table depicts the sizes subjected to Facebook upload from a desktop computer in order of smallest file dimension to the largest. The desktop dimension cutoff can be seen at 2048 with the line differentiating them and the difference in image width and height. The colored areas are the measured aspect ratios of each dimension. Highlighted in light pink, 1.333 is common for traditional television dimensions. Sizes colored in green represent 1.77 which is associated with High Definition video, 1.600 in red is linked to common computer screen ratios. Finally, colored yellow, 1.500 is typically found in Google Chromebook/ Microsoft Surface Pro 3 devices. As shown, images with an original width above 2048 pixels are resized by Facebook to 2048 pixels while retaining the original aspect ratio.
28


Table 4.5: Resized Images Spreadsheet with Aspect Ratios
UP Width UP Height Ratio DL Width DL Height Diff in Width Diff in Height
640 480 1.333 640 480 0 0
768 480 1.600 768 480 0 0
800 600 1.333 800 600 0 0
832 624 1.333 832 624 0 0
854 480 1.779 854 480 0 0
960 540 1.778 960 540 0 0
960 640 1.500 960 640 0 0
960 720 1.333 960 720 0 0
1024 576 1.778 1024 576 0 0
1024 640 1.600 1024 640 0 0
1024 768 1.333 1024 768 0 0
1136 640 1.775 1136 640 0 0
1152 720 1.600 1152 720 0 0
1152 768 1.500 1152 768 0 0
1152 864 1.333 1152 864 0 0
1280 720 1.778 1280 720 0 0
1280 800 1.600 1280 800 0 0
1280 960 1.333 1280 960 0 0
1334 750 1.779 1334 750 0 0
1366 768 1.779 1366 768 0 0
1400 1050 1.333 1400 1050 0 0
1440 900 1.600 1440 960 0 -60
1440 960 1.500 1440 960 0 0
1440 1080 1.333 1440 1080 0 0
1600 900 1.778 1600 900 0 0
1600 1200 1.333 1600 1200 0 0
1680 1050 1.600 1680 1050 0 0
1776 1000 1.776 1776 1000 0 0
1792 1344 1.333 1792 1344 0 0
1856 1392 1.333 1856 1392 0 0
1920 1080 1.778 1920 1080 0 0
1920 1200 1.600 1920 1200 0 0
1920 1280 1.500 1920 1280 0 0
1920 1440 1.333 1920 1440 0 0
2048 1152 1.778 2048 1152 0 0
2048 1280 1.600 2048 1280 0 0
2048 1536 1.333 2048 1536 0 0
2160 1440 1.500 2048 1365 -112 -75
2304 1440 1.600 2048 1280 -256 -160
29


UP Width UP Height Ratio DL Width DL Height Diff in Width Diff in Height
2304 1728 1.333 2048 1536 -256 -192
2560 1440 1.778 2048 1152 -512 -288
2560 1600 1.600 2048 1280 -512 -320
2560 1700 1.506 2048 1360 -512 -340
2560 1920 1.333 2048 1536 -512 -384
2736 1824 1.500 2048 1365 -688 -459
2800 2100 1.333 2048 1536 -752 -564
2880 1800 1.600 2048 1280 -832 -520
3000 2000 1.500 2048 1365 -952 -635
3200 1800 1.778 2048 1152 -1152 -648
3200 2400 1.333 2048 1536 -1152 -864
3840 2160 1.778 2048 1152 -1792 -1008
3840 2400 1.600 2048 1280 -1792 -1120
4096 2304 1.778 2048 1152 -2048 -1152
4096 3072 1.333 2048 1536 -2048 -1536
5120 2880 1.778 2048 1152 -3072 -1728
5120 3200 1.600 2048 1280 -3072 -1920
Table 4.5: Resized Images Spreadsheet with Aspect Ratios, Cont'd
TEST TWO RESULTS: MOBILE RESIZED
Once the desktop analysis was completed, a smaller group of sizes was selected to process through a mobile device. 12 different sizes were uploaded to Facebook. Table 4.6, below shows the same information as above but these are resulting sizes from the mobile processing. This table also shows the cutoff dimension of 1440 and the aspect ratios that were consistent with those mentioned above. Most these ratios are not common desktop or television monitors, which is consistent with a mobile processing.
30


Table 4.6: Aspect Ratios for Mobile Uploaded Resized Image Sizes
Mobile Upload Size Aspect Download Size Diff in Width Diff in Height
640 480 1.333 640 480 0 0
800 480 1.667 800 480 0 0
960 540 1.818 960 528 0 -12
1024 1024 1.000 1024 1024 0 0
1024 768 1.333 1024 768 0 0
1152 768 1.500 1152 768 0 0
1400 1050 1.346 1400 1040 0 -10
1600 900 1.800 1440 800 -160 -100
2304 1440 1.607 1440 896 -864 -544
2800 2100 1.343 1440 1072 -1360 -1028
4096 2304 1.800 1440 800 -2656 -1504
5120 3200 1.607 1440 896 -3680 -2304
ADDITIONAL OBSERVATIONS
During this research, there was an additional observation related to how Facebook gives hosted images filenames. When an image is uploaded, either as a single image or as an album, Facebook removes any previous filename that is provided and allows you to input your own name. But, once the image is downloaded the given filename does not transfer. Instead, there is a string of numbers that accompanies the image. The following table shows the filenaming pattern, images that are smaller than 1024 pixels result in a filename ending in "_n" and files larger than 1024 end in "_o".
Table 4.7: Facebook Filename with Ending Variables and Image Dimensions
Facebook Filename File Height File Width
14606261_171049506682465_7659861344040417242_n 640 480
14725616_171051096682306_7490145527690776057_n 640 512
14705798_171060090014740_47366085576596930_n 768 480
14716085_171062426681173_6167163997369377713_n 800 480
14708146_171066800014069_3999517150370107387_n 800 600
1466603l_171067920013957_8179131341331876227_n 832 624
14563588_171068510013898_9057441696503493076_n 848 480
31


14642395_171122626675153_3371192182059220701_n 854 480
14657278_171124130008336_4882842056698807867_n 960 540
14479817_171124653341617_2314511886844732288_n 960 544
14716200_171125273341555_6404543742962279837_n 960 640
14718609_171125593341523_780674529347943217_n 960 720
14691427_171126130008136_1274289947829898777_o 1024 576
14682179_171127633341319_2909700528195634371_o 1024 600
14714846_171128420007907_1536095147209096540_o 1024 640
14708047_171129080007841_5277984069627771666_o 1024 768
14753459_171130103341072_135969008382315551_o 1024 800
14700768_171130696674346_1405553320647029812_o 1024 1024
14700890_171131473340935_6017210988428949073_o 1080 1200
14682113_171132020007547_1541663541361619100_o 1120 832
14706862_171132540007495_2934281638814907577_o 1136 640
14711185_171133660007383_2636344962224353594_o 1152 720
14700997_171134123340670_5081340380920779830_o 1152 768
14753756_171135110007238_6992619365143712932_o 1152 864
14753914_171135916673824_9121976801749768309_o 1152 900
14633386_173871909733558_6675736714468206228_o 1280 720
14712478173872723066810122167938317306009_o 1280 768
14612406_173873519733397_6106772829617656390_o 1280 800
14753495_173878616399554_8584070896963507744_o 1280 854
14706791_173879643066118_5269638416668567012_o 1280 960
14682135_173880293066053_4437180080076288302_o 1280 1024
14589977_173880773066005_963340077428580666_o 1334 750
14712513_173881903065892_1269520307219485632_o 1366 768
14682149_173883026399113_829923068123835390_o 1400 1050
14713078174823156305100_7596506905553723884_o 1440 900
14681010_174824396304976_489873779676078849_o 1440 960
14712920_173887056398710_4551942151696058904_o 1440 1024
14691379_173887746398641_8957004741688044114_o 1440 1080
14615717_173888563065226_5473719157032050189_o 1600 768
14608784_173889006398515_8633918321371155614_o 1600 900
14691409_173890853064997_2722807948570474530_o 1600 1024
14633178_173891483064934_4747248546977408025_o 1600 1200
14706924_173892249731524_4476169264367838658_o 1600 1280
14615791_173893139731435_1185552378688184095_o 1680 1050
14691361_173893899731359_2854693465666411016_o 1776 1000
14691437_173894289731320_6627944137283334531_o 1792 1344
14681993_173894756397940_3430965673716753687_o 1800 1440
14711064_173895256397890_6065145293086472500_o 1856 1392
32


14682179_173895686397847_378153592677772147_o 1920 1080
14753718_173896463064436_975367263671377477_o 1920 1200
14707827_173897933064289_3862273431261348758_o 1920 1280
14711107_173898956397520_9197608778520528756_o 1920 1400
14714860_173899636397452_5940471284363883874_o 1920 1440
14543703_173900363064046_2316545056323288976_o 2048 1152
14753223_173908293063253_220597512020849299_o 2048 1280
14753381_173909553063127_5124466290378692774_o 2048 1536
14712572173910729729676853857490039576813l_o 2160 1440
14692105_173911283062954_1274641796107498291_o 2304 1440
14707010_173913263062756_1622449704346333317_o 2304 1728
14692122_173913569729392_6670188716681293260_o 2538 1080
14589953_173914149729334_6956274026427452684_o 2560 1080
14713093_173915099729239_6481259652049035234_o 2560 1440
14753348173915693062513_1524924099733716226_o 2560 1600
1463320417391617972913l_1342697296373069749_o 2560 1700
14680867_173917016395714_6401213935393268669_o 2560 1800
14753197_173917573062325_4654743316038574176_o 2560 1920
14633595_173918073062275_1049286623655939816_o 2560 2048
14712767173943136393102_2195393490275745802_o 2732 2048
14608763173943793059703511838646661087015_o 2736 1824
14706981_173945643059518_4401544997433701240_o 2800 2100
146919361739463930594433837396610021786855_o 2880 900
146821711739470397260452056195719809512504_o 2880 1800
14691291_173947849725964_5939741390186389714_o 3000 2000
14714957_173948813059201_1088149077070216294_o 3200 1800
14712970_173949323059150_2618835148265976661_o 3200 2048
14680862_173949796392436_2118268057472328063_o 3200 2400
14708006_173951626392253_2393911118619360305_o 3440 1440
14691943_173952683058814_3415801008302557676_o 3840 2160
14712627_173954316391984_2679064586028443444_o 3840 2400
14707781_173954983058584_6151962065799383493_o 4096 2304
14567464_173955546391861_7288819382589088815_o 4096 3072
14715026_173955946391821_3963479491714816446_o 5120 2160
14713086_173956346391781_5716767949075741631_o 5120 2880
14615727_173957133058369_5471907928923142878_o 5120 3200
Table 4.7: Facebook Filename with Ending Variables and Image Dimensions, Cont'd
33


CHAPTER V
DISCUSSION
The results from the first test have revealed that regardless of an image being uploaded one at a time or in an album, the observed variables are affected in unique ways based on the following three upload and download combination scenarios: 1) image is downloaded to a mobile device, 2) a mobile device uploads an image that is downloaded to a desktop computer, and 3) image is both uploaded and downloaded from desktop computers.
It was observed that images downloaded to a mobile device were Baseline encoded while images downloaded to a desktop computer were Progressively encoded. Progressive DCT is indicative of images that are loaded as full images and then progressively become clearer as the image moves from a low-resolution to a high-resolution image. Baseline DCT is typical encoding for smaller images and are loaded from top to bottom, pixel by pixel. These results from each schema are indicative of how the images were processed for viewing on various devices. Mobile devices have a smaller viewing window and are easier to load via Baseline DCT. In regards to Progressive DCT and its presence in remaining patterns, both were processed and downloaded via desktop. A desktop has a larger viewing area and can take slightly longer to load unlike a mobile. The image quality suffered slightly in the second schema, this is because the image was uploaded on a smaller screen and then forced to download to a larger viewing device.
34


After processing 56 different image dimensions it was concluded that the cutoff
image dimension of a picture uploaded to Facebookfrom a mobile device is 1440 pixels, while images uploaded from a desktop computer is 2048 pixels.
FUTURE RESEARCH
Some ideas for future research would be to examine the possibility of mobile album download when, and if, it becomes available to users. Although it did not show any signs of significance through this research it should still be examined for validity and embedded metadata analysis. This research was carried out on images with a portrait orientation. The same research should be performed on landscape images and determine any similarities and aspect ratio sizes. Facebook is a popular social media site, but there are several others that billions of people use daily. These sites should be explored in a similar manner and analyzed for embedded content. There may be similarities between this research and Instagram since it was bought by Facebook in April of 2012 [J], These sites are not all the same but they allow for the fundamental ability to share and download images.
35


REFERENCES
1. A brief history of Facebook | Technology | The Guardian, (n.d.). Retrieved October 11, 2016, from https://www.theguardian.com/technology/2007/jul/25/media.newmedia
2. Baseline JPEG compression juggles image quality and size | EE Times, (n.d.). Retrieved
November 28, 2016, from http://www.eetimes.com/document.asp?doc_id=1225736
3. Castiglione, A., Cattaneo, G., & Santis, A. D. (2011). A Forensic Analysis of Images on Online
Social Networks. In 2011 Third International Conference on Intelligent Networking and Collaborative Systems (INCoS) (pp. 679-684). https://doi.org/10.1109/INCoS.2011.17
4. Company Info | Facebook Newsroom, (n.d.). Retrieved from
http://newsroom .f b.com/com pa ny-i nfo/
5. Deever, A., Kumar, M., & Pillman, B. (2013). Digital Camera Image Formation: Processing and
Storage. In H. T. Senear & N. Memon (Eds.), Digital Image Forensics (pp. 45-77). Springer New York, https://doi.org/10.1007/978-l-4614-0757-7_2
6. Fowler, G. A. (2012, October 5). Facebook: One Billion and Counting. Wall Street
Journal, Eastern Edition, p. B.l. New York, N.Y.
7. Liu, Y., Gummadi, K. P., Krishnamurthy, B., & Mislove, A. (2011). Analyzing Facebook Privacy
Settings: User Expectations vs. Reality. In Proceedings of the 2011 ACM SIGCOMM Conference on Internet Measurement Conference (pp. 61-70). New York, NY, USA: ACM. https://doi.org/10.1145/2068816.2068823
8. Marrion, C. (2016). Digital Image Manipulation Detection on Facebook Images. University of
Colorado at Denver.
9. Medialnfo. (n.d.). Retrieved November 29, 2016, from https://mediaarea.net/en/Medialnfo
10. Moroney, N., & Fairchild, M. D. (1995). Color space selection for JPEG image compression.
Journal of Electronic Imaging, 4(4), 373-381. https://doi.org/10.1117/12.217266
11. Noda, H., Takao, N., & Niimi, M. (2007). Colorization in YCbCr Space and its Application to
Improve Quality of JPEG Color Images. In 2007 IEEE International Conference on Image Processing (Vol. 4, p. IV-385-IV-388). https://doi.org/10.1109/ICIP.2007.4380035
12. Rusli, E. M. (1333991748). Facebook Buys Instagram for $1 Billion. Retrieved October 3, 2016,
from http://dealbook.nytimes.com/2012/04/09/facebook-buys-instagram-for-l-billion/
13.SWGDE. (n.d.). Retrieved October 23, 2016, from https://www.swgde.org/documents
36


14.SWGIT. (n.d.). Retrieved December 1, 2016, from https://www.swgit.org/documents
15.Wu, H., Sun, X., Yang, J., Zeng, W., & Wu, F. (2016). Lossless Compression of JPEG Coded Photo Collections. IEEE Transactions on Image Processing, 25(6), 2684-2696. https://doi.org/10.1109/TIP.2016.2551366
37


Full Text

PAGE 1

nrn rrr !"! #!$%&'() #*+ # %-*. !" #%,*% -#!!.!" # !"!.!%,!/% %."*."..+ !" #0*+ "! #,$!" % !"-n-!,$ &'(1

PAGE 2

# #"! #% !"--,$ $%r// #%%//!,"! # n-!,$ r!$%+ % %.$!%#% 2""+ # 2%!3 -+(4 #&'(1

PAGE 3

r//$%5n-!,$ 6$ %.+%$n-!+/!%.%-!!7#,, % r!"!% %.$ !% !-%.+,%#%-!+-!/!% !!*,%. !* %,#%-#%$,#!3 3-!%,-!++*-% 3 #! #/!/.# !"".$# ,""%//%%-%,"*!%. #%/+%. "!-*,!#%$+%$ 3,",8# # #+%$%!"/!/ ./.%-! #$ # !".!!, %!-%. %,/! !*/.!%,$ #-%-! % %.%$%+!* !"+,,,,% %# #%+ !, "! # #""!"+!.% ,,7 !/*/.!%,$!+%$%3..% #""!+%$,+!#-#%%-%-#%$, ##! %-!!7 #-%%,3..%%.9, 3 !,"" # #:/.!/!*%-# !!-%. +,% %,2r-!+/! 3#./!,$%!%...!!7% %-!!7: #"!*,% !%. /!",$ %.+%$ -!+/!3../ ,#%/ ; #/ !-!",% %-!..!%,//%% "! #%-#,-,#*. "!..!3%,.%!% #",$!" # %.."* *%-# !/-%/* /!,"!"* #:%+% #"!+%,-! !" #% %%%//!, -!++, /*.-% //!,% %.$!%

PAGE 4

nn 3!*.,.7 #%7+/% "! #-! % / *# "!+!!* !"." %, !%--/ ,"% #%7!*"! %-#$ + #%.*!"%,!..%%,%,% ,!..%/ !*! ##!3,+#!3 !3!7"!3# % !*3% !* !"."%70* %, !.%"!+!*+ %7#%7!*"!-! *%..$+$$ %,.!*, -#.%, 2* 3% #%7!*"!!*,%.*3% $*//! !*$%+ +! % !%,,!3# 3%,,! !!7%$% ,%.!"+3!%, #!*$#!* #,%!%,#%,., 3 # %,:/-$,%3 # !*+,#% #$% $" -%! 3% !3#!*." !$ # %7 *%,!*3.."!+"%! /!!% #.!!*

PAGE 5

r 3% #%72""<% %."! #*//! *$$ !%,++,% /! !++%.#%.%,+! # % #!*$# /!. #!*$# # -,. %-#$#".,!",$ %."!*-#%!.$%%%,%+ #%7"*. "! #!//! $!3%+$% "*." #% !%. !",%!"!/!,$%3!,"*.!+ !.%*-# %:$,-/. %#!*!-7=!*%.3%/!,, !++%.3 #-.% %,/!+/ !*+%,-.%,%,.%,+/! % %.3.. >7!3%,!*3%.3%%%.%. "!+ #!*%,!"0* !!*+%,.%$% %, %#% +*-#% #%7 !* 3 #%7!*"!%$$ !%/% !"+ -!++ ?!,$ !"!*-../#!"!--.%%,%.*,!* !/!! #%-#!*% -,.7!3.,$%.%,%3!,"*. %-# #!/ #% -%-! !.%"!+ !* #!*$#!* +-%

PAGE 6

@rnnAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAA(rnBnn@AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAA& @rnBnB8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAA) AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAA4 @8nnrAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAC 2rBnB8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAD @rnrnrAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAA(& nrAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAA(1 rAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAA(4 B@rnnAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAA(C AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAA(C 8rnEAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAA&4 8nEAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAA&D nBAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAA&D B@rnAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAA;; nnn@AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAA;) nnAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAA;F

PAGE 7

n (( :" !!.* /* "!+$%.+%$AAAAAAAAAAAAAAAA AAAAAAAAAAAAA; ;(.%++%$+!G*% 9% !%. %,:" !!. %,% %*+/"! $%.+%$(5n-!+/! 6AAAAAAAAAAA AAAAAAAAAAAAAAAA(F )( n*. % :"!!.%,7 !/r!-$AAA AAAAAAAAAAAAAAAAA(4 )& !.!3.!%,,-#+%H%-!!7.%++%$ +!:" !!. %,% % *+/%,G*% 9% !%.AAAAAAAAAAAAAAAAAAAAAAA AAAAAAAA(C ); !./.!%,7 !/!3.!%,-#+%H%-!!7 .%++%$+! :" !!. %,% %*+/%,G*% 9% !%.AAAAA AAAAAAAAAAAAAAA&( )) 7 !//.!%,7 !/!3.!%,-#+%H%-!!7 .%++%$+! :" !!. %,% %*+/%,G*% 9% !%.AAAAA AAAAAAAAAAAAAAA&) )F n9,+%$/%,# 3 #/n% !AAAAAA AAAAAAAAAAAAAAAA&4 )1 /n% !"!!./.!%,,n9,+%$9 AAAAAAAAAAAAAAAAA;' )4 %-!!7.%+3 #,$B%%.%,+%$ +!AAAAAAAAAAA;(

PAGE 8

nn &( /$ %.+%$!+/!Hn #!*$# !,$AAAAAAAAAAAAAAAAA(( ;( % %% :AAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAA(1

PAGE 9

: nnn I!.!. %I!%"!+nI$ %.$.>n".:I!.%.rI.!%.r! !$ +2rI2! r#! !$%/#-:/ !*/nIn*$ #!,$8I"-8!7$!*/!$ %.,8I"-8!7$!*/!+%$$-#!. !$ GIG*%. %%$+ +

PAGE 10

n nn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

PAGE 11

& /!/.%, #!!",-/ ,,*%. #-% %--!+/.#,*$,$ %. "!!-,$ %.+%$%%.-%%,.!-% $ %, / $ #+,,, ,% %-!,,,!",$ %.+%$ #" -8!7$!*/!$ %.,586,"+*. +,%,-%J%.!$!, $ %.+,%-.*,$* .+ !".+ %/+%$ -%,!/ -%.+,%%,P! #"!+% !-! %, #KL(;M rnBnn@ ##%/!*%-# $% $ # "!-%%.!",$ %.+%$ -!+/!%, #+/%!-%.+,%%//.-% #%!/!-,+%$8* %. /!/!,%!.* !J-!+/% !"2r-!, ,-!.% ,+%$3 #!* .!K +!$ %, %/ !#,! ,-!,$%+%$-!..!%,% /,! **-!+/., #% -%, + %-!++!%. %-! #+%$ #!+.% -%:/.! ,"! #-!..!% # #%%-#,,*%./* # /!"%%.-%"-%.%,,*-! "! !%$!".%$!-%.+,% L(FM $%, !,"" !-%. 3!7$ % $.! %.%%.9,% #,"" !-%. &'((#-!..,%!* !"+%$ #% -.*,,,"" +%$ !.* !%,9%3..% #*+%.%,/! "./*#, +, #% -!+/!+!,Q+ %,% %"!+/!-,+ %$%,,"" 0*%. +%$ %..!3.+ ,/:.!.* !L;M

PAGE 12

; %!:/.!, #,-! %"!+5 6+%/%,!..%%.56 !"+%$/!-, #!*$#%-!!7 !, % ,-% !!"+%/*.% !# *. -!-.*,, #% !-% %+/,+%$3%-! +/,%-!!7 #/!* +%/*.% !3%**%..*, %.#-!+/ !%"", #*!*,$ $#!$/:.%,+%7, #%. % !LCM

PAGE 13

) nn n #+,,,,% % /-%..%!$%.+%$-% "!*, #"!..!3$ %.#%/% %.!* /* $% ,*$ :" !!.# %.#!3 #".9". /+%7%,+!,.!" #-%+% + %+/%, #$.!%./! !$ +5r6 "!+% %.((:" !!.* /* "!+$%.+%$:"!!.B!R('&F .%+R(D&':())'=?/$ !RPP+$%S'''P7 !/Pn9,+%$ .9R1CC)&C .!,"% R()41)FF'4';D;&1 .--% R()41)FF'4';CC&1 .% % R()41)FF'4';CC&1 .r+!R;;&'1 ./R2r ./: !R2r /R+%$P?/$ 2rrr'5() 6 TL%% %,!D M U2B!R(( Un!.* R( UQn!.* !R4& Un!.* !R4& 2rrr(5(&&C) 6 :" ,R TL',!3 #(( M U'6%7R//. U(6!,.Rr#!1 U&6 % !R1 U;6Qn!.* !R4&54&P(6 U)6n!.* !R4&54&P(6 UF6n!.* R& U16!" 3%RD;)

PAGE 14

F %.( ( :" !!. /* "!+$%.+%$ N, U46!,"% R&'(1'D'1'1)1'4 UC6r! !$R( UD6:""" 5*!6>>V UTL:",!3 #;& M UU'6:/!*+R''11111111145(P(F6 UU(6*+R&&5((PF6 UU&6:/!*r!$%+R& UU;6RF'' UU)6:"B!R'&&( UUF6% +$%.R&'(1'D'1'1)1'4 UU16% % R&'(1'D'1'1)1'4 UU46!+/! !"$*% !R(&;' UUC6#* /,B%.*R;D'1D1DC415(;&&DP;;C16 UUD6/ *B%.*R&&4F''4(&F54DC;P;F'D6 UU('6$# B%.*R>(C')&('4)5>(44;DPDC;&6 UU((6:/!*!+/% !R'5'P(6 UU(&6 $!,RF UU(;6.%#R(1 UU()6!-%.$ #R)(F5C;P&'6 UU(F6*?%R&'(F(F((&&(4(;;' UU(16%7! //.5*!6>>V UUTL%7! ,!3 #(( M UUU'6//.S':'''(R) UUU(6//.S':'''&R/. ''QMMMQnGnW70 MS-EBW":X:0WM-,MQ8MW7UL/M UUU&6n*+5*!6>>V UUUTL%r,!M UUUUn*+.%$R( UUUUn*+B%.*R)D')DFD1'F;(&F' UUUUn*+/!-#R' UUUUn*+-%.R(''''''''' UUU;6//.S':''')R( UUU)6//.S':'''FRC' UUUF6//.S':'''1RD& UUU16//.S':'''4R( UUU46//.S':'''CR'';4)&;')(;;>'44;(4;CF4 ;>'1';&('1'D5D;'P&)CF(> (C('P&;L/M UUUC6//.S':'''DR);4( UUUD6//.S':'''R' UUU('6//.S':''()R) UU(46*-+$%.R(;& UU(C6*-+$ 9,R(;&

PAGE 15

1 %.((:" !!.* /* "!+$%.+%$! N, UU(D6.%#/:B!R'('' UU&'6!.!/%-R( UU&(6:"+%$8, #R)';& UU&&6:"+%$@$# R;'&) UU&;6$ #!,R& UU&)6-/R( UU&F6:/!*!,R' UU&168# %.%-R' UU&46!-%.$ #;F++!+% R&D UU&C6-%/ */R' UU&D6"!R)(F)(F&&&&5C;P&'C;P&'( (PF((PF6 UU;'6%7R//. UU;(6!,.Rr#!1%-7-%+%)(F++"P && U('6r"!5*!6>>V UTLr,!3 #(F M UU'6r% *,n"R UU(6r% *,R;D;DF)&F5;DP(;DP(F)&FP( ''6 UU&6r!$ *,n"R8 UU;6r!$ *,R(')F1FC4C5(')P(F1P(FC4 CP(''6 UU)6r. *,n"R' UUF6r. *,R(1;'F')DF5(1)1C(P('(6 UU16r+ %+/R(&)115(&P()1P(1''P(''6 UU46r/,n"R UUC6r/,R'5'P(6 UUD6r+$!n"R UU('6r+$!R;&&&&FC'1F5DDCDP;(6 UU((6r %$n"R UU(&6r %$R;&&&&FC'1F5DDCDP;(6 UU(;6r% %+/R&'(1'D'1 UU()6r@r! !$!R('5('P(6 TL(,!3 #1 M U'6!+/!R1 U(6Qn!.* !R4&54&P(6 U&6n!.* !R4&54&P(6 U;6n!.* R& U)6#*+%."" R&'() UF6#*+%.$ #RD'(4 2rG51F 6 2rG51F 6 2r'5(F 6 +%$8, #R(D&'

PAGE 16

4 +%$@$# R())' -!,$r!-R' r%+/.RC !.!!+/! R; 2r@5&C 6 2r@54& 6 2r@5&1 6 2r@5)4 6 2r %.((:" !!.* /* "!+$%.+%$! N, #"!+% !-%*"*.% $% %,.,%,, !%.,% % #% -% %%.9,"!%* # /*/!# # / %-# !#!3 #+% %. -#%$,3#+%$%-!+/,%-!!7 +%.-* .,!! *% %,%,% %$3# -!+ !-% $"%. +%$../#!%-#%/ #%:/n 5$ %.$.>n".:6-%+% %, ../!,*-%#$#>!.* !+%$!+/* .%/ !/%,%/,/!-$ %. %%--. !%,!.$ %.+%$ %%,.%%.%.! # "! ",!3.!%,!"!/*-#% #!*$#-*-# %,! !-7#!.,""#+! % !%, !!" #/!3#!/!, *#%. ,+%$ "! *% .+%/*.% ,+%$#%%"",3. .>7!3/!/.*-#%/, /!. -%%,-. #+%$!%.!" # /* !73 #!%..+%$! !%-#!+ /!"/!. -%.$%#%. !-#%$%,$ %.+%$! %3 /#!+!* ,!0*, $!" 3%%, # %!.% # +> -!*+$7..#%%./!$%+"!/* -#%*-#%,!r#! !#!/%, !.r% #!/ #% %-!+/#%,!"" %,%-,"% *"!+%/*.%

PAGE 17

C /!$%+-%%.!,!3.!%,,"!+ # #% -%%--!+/.#+.%*. #!" 3%!/ !/%,3 #% %+!* !" J#!3> !K,!!!**%, "!+% !%."!*+%,-% $%+%/*.% ,+% $-%%--!+/.#,!+! 3 #!"!+%. %$ @8nnr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

PAGE 18

D +%$. #!*$#!3,"*83!7$$!*/ #%"!+,%,%%3 #8 2rBnB8 (DD& #2! r#! !$%/#-:/ !*/52r 6-!++ $% #, *, -!.!+%$%,,.!/ #3..>7!3 %,%," !+%$-!+/!L('M2r -!+/!.!3#-#8,"%J-!+/ !3#-#,% %.! %,-%! !$%."!+KL(;M2r%-# # *. !%%!"+%$ #% -! %.!3"0*--#%$ %,!" %7-!.! ,""-! #%-# # -!-3 #.!-!+/!%, ""! ,$ %.+%$/!-, #!*$# %-!!7 +/! % !+ #% #%+%,$ !"2r-!+/! #% -%%//., !+%$ #% *. %%$!" !%..0*%. ,$%,% !#%+!* !" 0*% 9% !!%,! #.!!"0*% 9% %.3#-#-%!*$#. %.% !%0*%. .. #% %$$"!+( 5 #.!3 0*%. 6 !DD5#$# 0*%. 6 /-%.. #%+!* !"/:."!+% #% !,! ,-%. # #*+%*.!*+!.!30*%. "%! @!3!-%".%,%, # -!+/!%.$! #+%//.,-! *!*.! %. % #%+!* !"*%. ,$%,% !3..-!++!.!..!3$ %$%.!3!" #2r -!+/!/!!.!-,"!+ #-!.!". %%56 -!++!.%n5,.* $6-!.!%.*%, #-! !" !-!+/!JK/ #

PAGE 19

(' 5.*+%6!.*+%-%.*3#-#+/. # !"$# %+%$JK%, JK/ # 3!-#!+%-%.*!".*% ,,#-!.!/%-+!%. :/.! ,"!-!+/!%,-% !,3 #% #$#+%$!.* !%, % ,% /%% "!$% ""--L((M $*&(,/#/!#% "!..!3!!#!:#!3 /!$! !%+%$$/. !C:C.!7!"/:.#!%%% # *?, !%!%"!+563##J-! /:.%.* %"!+ -!""-!/!,$ !/% %."0*-K LFM %//., #%.*% %-#/:.% 0*% 9,H #*. $%.*-% -!.% #3$# %-#/:./!, /!,*-$, %.!" #!%..+%$ # /-!+/ #+%$+!$ #. $"-% .! -%.-#%$ %.* 3/:.%, #+!/:. #% -% +%,*"!+ #+%.. # *. $".#*. +!$#$#"0* -, %.H%-#%$3#-#. -%. ##*+%"!+% !+!,,* $ #0*% 9% !/#%-%! ,,*$,-!+/!%,-! +! $"-% !2r-!+/!N .!L&M #/:.!3, !%%$,"!+%C:C. !-7 !%!>,+!%. %%"!*0* n*$ #!,$5n6%,@* ""+%-!,$#%-#, -!..$ #%.* % $% #*//." -! !" #.!-7%,-% $%9$>9%$ /% %.. #!*$# #.!-7* #"%./! %,"!+ #! !+$# -!

PAGE 20

(( $*&( /$ %.+%$!+/!Hn #!*$#!,$

PAGE 21

(& nnn n #%-# %-!,$+ #!,%//., !+ %$ #% %/!-, #!*$# %-!!73%%.9, 3!,"" 9$! #,7 !/%,+!.,# %-!,$ /!.-!+/!!" # +%$%, #-% !!"3 + %,% % #*. $2r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

PAGE 22

(; %.;(.%++%$+!G*% 9% %.%,:" !!. %,% % *+/"!$%.+%$(5n-!+/! 6%-!!7.%+&'(1'D'1S(&((F)?/$ +%$+!&DCC:F;(& :"!!.B!R('&F .%+R&'(1'D'1S(&((F)?/$ !RPP+$%S'''P7 !/P$%. +%$( .9RF'(FF&C .!,"% R()4;(4C1D4D)C;( .--% R()C'F1&)41;)&)F .% % R()C'F1&)41;)&)F .r+!R;;&'1 ./R2r ./: !R2r /R+%$P?/$ 2rrr(5(FDD' 6 :" ,R TL',!3 #(; M U'6+%$8, #RF;(& U(6+%$@$# R&DCC U&6%7R%+*$ U;6!,.R>>D&' U)6 % !R1 UF6Qn!.* !R4&54&P(6 U16n!.* !R4&54&P(6 U46n!.* R& UC6!" 3%RD&';r@) UD6!,"% R&'(1'D'1(&((F) U('6r! !$R( U((6:""" 5*!6>>V UTL:",!3 #&1 M UU'6:/!*+R'(5(P('6 UU(6*+R(D5(DP('6 UU&6:/!*r!$%+R& UU;6R&F' UU)6:"B!R'&&' UUF6% +$%.R&'(1'D'1(&((F) UU16% % R&'(1'D'1(&((F) UU46#* /,B%.*R;;&5;;&P(''6

PAGE 23

() %.;(.%++%$+!G*% 9% %.%, :" !!. %,% %*+/"!$%.+%$(5 n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r"!5*!6>>V UTLr,!3 #D M UU'6rB!R&&'' UU(6r% *,n"R UU&6r% *,R;)1;(5;)P(1P(;(P(6 UU;6r!$ *,n"R8 UU)6r!$ *,RC;F&;;5C;P(F&P(;;P(6 UUF6r. *,n"R( UU16r. *,R'5'P(6 UU46r+ %+/R(1((;C5(1P(((P(;CP(6 UUC6r% %+/R&'(1'D'1 TL(,!3 #D M U'6+%$8, #RF(&

PAGE 24

(F %.;(.%++%$+!G*% 9% %.%, :" !!. %,% %*+/"!$%.+%$(5 n-!+/! 6! N, U(6+%$@$# R&CC U&6!+/!R1 U;6 % !R1 U)6Qn!.* !R4&54&P(6 UF6n!.* !R4&54&P(6 U16n!.* R& U46#*+%."" R((F) UC6#*+%.$ #R()C;' 2rrr)5(;()& 6 2rrrF5&) 6 2r@5)(1 6 2rG5(;' 6 2r'5(F 6 +%$8, #RF;(& +%$@$# R&DCC -!,$r!-R' r%+/.RC !.!!+/! R; 2r %+*$ %.514 % !"" ':)-C4%F6 %+*$S%.S':'%'(>%+R+%$SS% % + %+/R()4;(4C;(F(;D ;))));;) ))));;;) ))));;)F );;;;)F1 ;;;;F14C ;;;)1CD(& ;;)F4D(&(F ))F1C(&(F&( ;;;;;;;) ;;;;;;;) ;;;;;;)F ;;;;;)F1

PAGE 25

(1 ;;;;F14C ;;;)1CD(& ;;)F4D(&(F ))F1C(&(F&( %.;(.%++%$+!G*% 9% %.%,:" !!. %,% % *+/"!$%.+%$(5n-!+/! 6 N, nr #% 3!,"" 3% !*/.!%,% +%$ #!*$#%-!!7%% ,7 !/3 #% !3%,%+!.,# #!*$#%!3!3 # # %-!!7%//.-% !8 # #!/ !%+%$ -%*/.!%,,%%$.+%$! %-!..!!"+%$-%..,%%.*+!3.!%, $-%/%. %.$# .+!.+ ,H +%$-%,!3.!%,,*$! #%+!.%,, 7 !/* % # +!" #%-# #%. !,!3.!%,%%.*+ #!*$#%+!., -! %%.%.# !/ !%..!3"! 3.,"" -%!%+% :3%-% !/* %..,% %%, /!/.:/.!%..!/ #% $*;(.!3

PAGE 26

(4 $*;( % %% : 8#.-!,$ #%*%!"!$%. +%$ #%+/! % % !3! %,"!+ #-../#! *, #" #+,,, ,% %"!+%-#/#! !3% ,"!-!+/%!% $% #/!-,+%$% "!.,3-% !/ %-#+%$%,% -#/+* % !!"+!.%,,7 !/ */.!%,$%,,!3.!%,$3$ #!3*" !.,#3%*, !7/%.. %,,% %/%% "!%%. #% 3...% :/.%,#%/ ) r #-!, 3%-!+/!,!" %7$%$.+% $ #% !$% ,"!+% //.r#!1%,-%.$ !F1-!++!+% $,+!#9%$, "!+ #+%.. 1)':)C' #.%$ F(&' :;&''%-#+%$3%*/.!%,,! #

PAGE 27

(C "!+%,7 !/%,+!.,-%,,!3.!%,, %,7 !/-!+/* !! # /% %-!!7* !9*/.!%,,+%$

PAGE 28

(D n %%.9$ #,% %"!+ #" 3% +, #% #! ,""--#%$ !%+%$ 3*/.!%,$ ,,*%..!%%.*+# +%,% !!--* #!*$# #+%-#+%/ #!*$#!* #"+%$# /% #% +!.,!3.!%,,+%$$ %,.!"#!3 #%*/.!%,,% -!+/,3 #/-"-0*% 9% %.* $%.-!,$%,/!,*-% /% ,+ %,% % **-!+/!,! "1F.+ #! # 3!-#+% .% !+%$,!3.!%,, !%,7 !/-!+/* #" !" # #-!,!%.. -#+%3%+%$ #% 3*/.!%,,"!+ %+!.,#,!3.!%,, %,7 !/-!+/* H #%-!+/,3 #, 0*% 9% %.*$ r!$-!,$%,%+ %,% % **!"14 .+ %..+%$3 */.!%,,%,,!3.!%,,"!+%,7 !/-!+/* % r!$-!,,%3..* #%%%.0*% 9% %.!*$#.0*%. !%2r0*%. "%!!"C4%.!$ 3 #%/% ,+ %,% % **!"C(.+ *++%!" #-%"!*, %.)(.!3#90*% 9% %.%, + %,% %,*+/"!% $% +%$5 #%+/ ,#%/ ;/%$(&%! 6%-#!" # #-%! "!..!3

PAGE 29

&' %.)(n*. % :"!!.%,7 !/r !-$ %-#!" #"!..!3$ %.%.. #-#+%% #/".%+ +%$,+!0*% 9% %.%,:" !!. %,% %,*+/#" %.#!3 #+!.,!3.!%,,+%$ #-!,,/# ,7 !/,!3.!%,,!* /* %, # "%.+%$3..,/.% #,7 !/,!3.!%,,/% %.)&!.!3.!%,,-#+%H%-!!7. %++%$+!:" !!. %,% %*+/%,G*% 9% !%.%-!!7.%+()&;C;''S(;4(F4D(;)')DFCSFFD;4;& ;D)FC1C'4)))S?/$ +%$+!F&C:D1' :"!!.B!R('&F .%+R()&;C;''S(;4(F4D(;)')DFCSFFD;4;&;D)FC1 C'4)))S?/$ !RPP+$%S'''P7 !/P!./. !%,!.!3.!%, .9R()4D)1 .!,"% R()4;F)(F;FF4DFD .--% R()C'F1)C(4)1441 .% % R()C'F1)C(4)1441 .r+!R;;&'1 ./R2r ./: !R2r

PAGE 30

&( %.)&!.!3.!%,,-#+%H%-!!7. %++%$ +!:" !!. %,% %*+/%,G*% 9% %.! N, /R+%$P?/$ 2rrr&5F;C 6 Sr!".-#*7(!"( TLSr!".,!3 #(' F&) M Ur!".@%,5*!6>>V UTL%% %,!(&C M UUr!"./R.-+ UUr!".B!RF&C UUr!"..%R+ UU!.!/%-% %Rn UUr!".!!/%-RQE UUr!".% +R&'(&(&F;)(F4 UUr!"..$% *R%-/ UUr+%r.% "!+Rrr UU.%$R' UU-%*"%*R UU-!,.R UU* R'' UUn,$ R' UU!!/%-..*+% R'D1)&('C&)D( UUr!".% !R.-+ UUr!".R'''''''''''''''' U'6r!".-/ !R-& U(6r!".!/$# R U&6,%8# r! R'D1)&('C&)D( U;6,%.%-7r! R''(&'F''(&F''(';( U)6n,% :!.*+R');1'4'&&&)D''(;D& UF6% :!.*+R';CF(F'4(1C4''D4'C U16.*% :!.*+R'();'4''1'1('4()( U46n,nR-*-7YG)=6&HG3 M7/9UZ' UC6nR-*-7YG)=6&H G3M7/9UZ' UD6.*nR-*-7YG)=6&HG 3M7/9UZ' 2rrr'5() 6 TL%% %,!D M U2B!R(( Un!.* R' UQn!.* !R( Un!.* !R( 2rrr(;5F& 6 TLr#! !#!/,!;C M

PAGE 31

&& %.)&!.!3.!%,,-#+%H%-!!7. %++%$ +!:" !!. %,% %*+/%,G*% 9% %.! N, Ur% %5*!6>>V UTLr,!&F M UU* r$ RB8 UU>>r//.-% !-!,>> UU$%.%+!n"-R,C.#CBC+ @QG 2rG51F 6 2rG51F 6 2r&5(F 6 +%$8, #RF&C +%$@$# RD1' -!,$r!-R& r%+/.RC !.!!+/! R; 2r@5&1 6 2r@5&F 6 2r ;&;;)F('() &&;;)4(;(C &;;)4(((1(D ;)F1(((;(4&' FFC('()(1&(&& C(&(((4&&&(&)&' ('(&()(1&(&;&)&( (&(((((&(F(C&'&' ;)FD&'&'&'&' ))F(;&'&'&'&' FF((&'&'&'&'&' D(;&'&'&'&'&'&' &'&'&'&'&'&'&'&' &'&'&'&'&'&'&'&' &'&'&'&'&'&'&'&' &'&'&'&'&'&'&'&' %.)&!.!3.!%,,-#+%H%-!!7. %++%$+!:" !!. %,% %*+/%,G*% 9% !%.! N,

PAGE 32

&; %.);!./.!%,7 !/!3.!%,-#+%H %-!!7.%++%$ +!:" !!. %,% %*+/%,G*% 9% %.%-!!7.%+()(DD&'CS(;&&4411;CD&DC;S1')C1C4 F(FC(4FC);1FS5(6 +%$+!F&C:D1' :"!!.B!R('&F .%+R()(DD&'CS(;&&4411;CD&DC;S1')C1C4F(FC(4 FC);1FS5(6?/$ !RPP+$%S'''P7 !/P!./. !%,7 !/!3.!%, .9R()4D)1 .!,"% R()4;;F&C(&('&CC .--% R()C'F1;D4'&)'); .% % R()C'F1;D4'&)'); .r+!R;;&'1 ./R2r ./: !R2r /R+%$P?/$ 2rrr&5F;C 6 Sr!".-#*7(!"( TLSr!".,!3 #(' F&) M Ur!".@%,5*!6>>V UTL%% %,!(&C M UUr!"./R.-+ UUr!".B!RF&C UUr!"..%R+ UU!.!/%-% %Rn UUr!".!!/%-RQE UUr!".% +R&'(&(&F;)(F4 UUr!"..$% *R%-/ UUr+%r.% "!+Rrr UU.%$R' UU-%*"%*R UU-!,.R UU* R'' UUn,$ R' UU!!/%-..*+% R'D1)&('C&)D( UUr!".% !R.-+ UUr!".R'''''''''''''''' U'6r!".-/ !R-& U(6r!".!/$# R U&6,%8# r! R'D1)&('C&)D(

PAGE 33

&) %.);!./.!%,7 !/!3.!%,-#+%H %-!!7 .%++%$+!:" !!. %,% %*+/% ,G*% 9% %.! N, U;6,%.%-7r! R''(&'F''(&F''(';( U)6n,% :!.*+R');1'4'&&&)D''(;D& UF6% :!.*+R';CF(F'4(1C4''D4'C U16.*% :!.*+R'();'4''1'1('4()( U46n,nR-*-7YG)=6&HG3 M7/9UZ' UC6nR-*-7YG)=6&H G3M7/9UZ' UD6.*nR-*-7YG)=6&HG 3M7/9UZ' 2rrr'5() 6 TL%% %,!D M U2B!R(( Un!.* R' UQn!.* !R( Un!.* !R( 2rrr(;5F& 6 TLr#! !#!/,!;C M Ur% %5*!6>>V UTLr,!&F M UU* r$ RCWW9:%; UU>>r//.-% !-!,>> UU$%.%+!n"-R*%+3!@*9 -(CF>#% 2rG51F 6 2rG51F 6 2r&5(F 6 +%$8, #RF&C +%$@$# RD1' -!,$r!-R& r%+/.RC !.!!+/! R; 2r@5&1 6 2r@5&F 6 2r

PAGE 34

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r+!R;;&'1 ./R2r ./: !R2r /R+%$P?/$ 2rrr'5() 6

PAGE 35

&1 %.))7 !//.!%,7 !/!3.!%,-#+%H %-!!7.%+ +%$+!:" !!. %,% %*+/%,G*% 9 % !%.! N, TL%% %,!D M U2B!R(& Un!.* R' UQn!.* !R( Un!.* !R( 2rrr(;5(F) 6 TLr#! !#!/,!()' M Ur% %5*!6>>V UTLr,!(&C M UU* r$ R>nP"Y[ UU>>r//.-% !-!,>> UU$%.%+!n"-R)'"1&&! ($r;r'' UU/-%. *!R'('''%D',''''&C&''''D&'(''1;'(''%-)'(''4 F,"'('''L/M 2rrr&5;'1& 6 Sr!".-#*7(!"( TLSr!".,!3 #(1 ;')C M Ur!".@%,5*!6>>V UTL%% %,!(&C M UUr!"./R UUr!".B!RF(& UUr!"..%R+ UU!.!/%-% %Rn UUr!".!!/%-RQE UUr!".% +R&''D;&4&(;1;( UUr!"..$% *R%-/ UUr+%r.% "!+R UU.%$R' UU-%*"%*R UU-!,.R UU* R(' UUn,$ R' UU!!/%-..*+% R'D1)&('C&)D( UUr!".% !R UUr!".R)(&)C1(&&&(4F&)&CF(4)(&' 11&F'&&C&'&(;(F4(; U'6r!".-/ !Rn1(D11>&>(.%7-%., U(6.*% :!.*+R'();'4''1'1('4()( U&6.*nR-*[5>&4H\2W-#+3U OL/M U;6-!,.-R1(D11>&>("%*. n !.!*/%-I$ U)6% :!.*+R';CF(F'4(1C4''D4'C

PAGE 36

&4 %.))7 !//.!%,7 !/!3.!%,-#+%H %-!!7.%+ +%$+!:" !!. %,% %*+/%,G*% 9 % !%.! N, UF6nR-*[5>&4H\2W-#+3U L/M U16*+%-R'C'' U46%*+ 5*!6>>V UTL%% %,!;1 M UU%*+ R( UU%*+ %-7$R''' UU%*+ !+ R' UU%*+ .%R' UU%*+ ..*+% R& UC6,%.%-7r! R''(&'F''(&F''(';( UD6n,% :!.*+R');1'4'&&&)D''(;D& U('6n,nR-*[5>&4H\2W-#+3U OL/M U((6-#!.!$Rn U(&6B3$!,-Rn"-B3$!, 1(D11>&>( U(;6,%8# r! R'D1)&('C&)D( U()6r!".!/$# R!/$# % !%. !.!!! *+&''D U(F6#!+% -,%/ % !R(')4D(''&&D;>''F '&''&D1'DD')1>''(4'4>'''D&F L/M 2rG51F 6 2rG51F 6 2r&5(F 6 +%$8, #R((F& +%$@$# R&')C -!,$r!-R& r%+/.RC !.!!+/! R; 2r@5&F 6 2r@5&) 6 2r@5&) 6 2r 1))1('(1&'&) FF1C('&;&)&& 1F1('(1&;&C&& 14D(&&';F;&&F 4D(F&&&4)))(;( ('()&&&1;&)&)F;4 &'&1;(;F)()C)C)' &D;4;C;D)F)')()'

PAGE 37

&C 44('(D)')')')' 4C('&1)')')')' ('('&&)')')')')' (D&1)')')')')')' )')')')')')')')' )')')')')')')')' )')')')')')')')' )')')')')')')')' %.))7 !//.!%,7 !/!3.!%,-#+%H %-!!7.%++%$ +!:" !!. %,% %*+/%,G*% 9% %.! N, 8nrnE :%+$ #+%$9,%-!!7 3%, +, #/% -*.% #% #-%!3 # #,# ,+!-* !"""!+!./* 3, #())'/:.%,,7 !//*3, #& ')C/:.#"!..!3$ %.,/#9*?, !%-!!7*/.!%,"!+%,7 !/-!+/* !,!"+%.. ". ,+! #.%$ #,7 !/,+!-* !""-%% &')C3 # #. ,"" % $ #+%, #,""-+%$3 #%,#$# #-!.!,%%% #+%*,%/% !!"%-#,+!@$#. $# ,.$# /7(;;;-!++! "! %, !%. .!,+!9-!.! ,$/ (443#-# %!-% ,3 #@$#" !,!(1'', .7, !-!++!-!+/* % !%..-!.!,..!3(F'' /-%.. "!*,!!$.#!+!!7P-!!" *"%-r!;,-#!3+%$3 #%! $%.3, #%!&')C/:.% 9,%-!!7 !&')C/:.3#. %$ #!$%.%/%

PAGE 38

&D %.)Fn9,+%$/%,# 3 #/n % 1)')C' (;;; 1)')C''' 41C)C' (1'' 41C)C''' C''1'' (;;; C''1'''' C;&1&) (;;; C;&1&)'' CF))C' (44D CF))C''' D1'F)' (44C D1'F)''' D1'1)' (F'' D1'1)''' D1'4&' (;;; D1'4&''' ('&)F41 (44C ('&)F41'' ('&)1)' (1'' ('&)1)''' ('&)41C (;;; ('&)41C'' ((;11)' (44F ((;11)''' ((F&4&' (1'' ((F&4&''' ((F&41C (F'' ((F&41C'' ((F&C1) (;;; ((F&C1)'' (&C'4&' (44C (&C'4&''' (&C'C'' (1'' (&C'C'''' (&C'D1' (;;; (&C'D1''' (;;)4F' (44D (;;)4F''' (;1141C (44D (;1141C'' ()''('F' (;;; ()''('F''' ())'D'' (1'' ())'D1''>1' ())'D1' (F'' ())'D1''' ())'('C' (;;; ())'('C''' (1''D'' (44C (1''D'''' (1''(&'' (;;; (1''(&'''' (1C'('F' (1'' (1C'('F''' (441(''' (441 (441(''''' (4D&(;)) (;;; (4D&(;))'' (CF1(;D& (;;; (CF1(;D&'' (D&'('C' (44C (D&'('C''' (D&'(&'' (1'' (D&'(&'''' (D&' (&C' (F'' (D&' (&C' (D&'())' (;;; (D&'())''' &')C((F& (44C &')C ((F&'' &')C(&C' (1'' &')C (&C''' &')C(F;1 (;;; &')C (F;1'' &(1'())' (F'' &')C (;1F>((&>4F &;') ())' (1'' &')C (&C' >&F1 >(1'

PAGE 39

;' &;')(4&C (;;; &')C (F;1>&F1>(D& &F1'())' (44C &')C ((F&>F(&>&CC &F1'(1'' (1'' &')C (&C'>F(&>;&' &F1'(4'' (F'1 &')C (;1'>F(&>;)' &F1'(D&' (;;; &')C (F;1>F(&>;C) &4;1(C&) (F'' &')C (;1F>1CC>)FD &C''&('' (;;; &')C (F;1>4F&>F1) &CC'(C'' (1'' &')C (&C'>C;&>F&' ;'''&''' (F'' &')C (;1F>DF&>1;F ;&''(C'' (44C &')C ((F&>((F&>1)C ;&''&)'' (;;; &')C (F;1>((F&>C1) ;C)'&(1' (44C &')C ((F&>(4D&>(''C ;C)'&)'' (1'' &')C (&C'>(4D&>((&' )'D1&;') (44C &')C ((F&>&')C>((F& )'D1;'4& (;;; &')C (F;1>&')C>(F;1 F(&'&CC' (44C &')C ((F&>;'4&>(4&C F(&';&'' (1'' &')C (&C'>;'4&>(D&' %.)Fn9,+%$/%,# 3 #/n % !! N, 8nnE #,7 !/%%.3%-!+/. ,%+%.. $!*/!"93%., /!#!*$#%+!.,-(&,"" 9 3*/.!%,, !%-!!7%.)1 .!3#!3 #%+"!+% !%%!* # %*. $9"!+ #+!. /!-$# %.%.!#!3 #-* !"",+ !!"())'%, #%/% #% 3-! 3 # #!+ !,%!! # % !%! -!++!,7 !/ .!+! !3#-#-! 3 #% +!./!-$

PAGE 40

;( %.)1/n% !"!!./.!%,,n9 ,+%$9 n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

PAGE 41

;& ()1)&;DFS(4((&&1&114F(F;S;;4((D&(C&'FD&&'4'(SCF) )C' ()1F4&4CS(4((&)(;'''C;;1S)CC&C)&'F11DCC'4C14SD1' F)' ())4DC(4S(4((&)1F;;)(1(4S&;()F((CC1C))4;&&CCSD1' F)) ()4(1&''S(4((&F&4;;)(FFFS1)')F);4)&D1&&4DC;4SD1' 1)' ()4(C1'DS(4((&FFD;;)(F&;S4C'14)F&D;)4D);&(4SD1'4 &' ()1D()&4S(4((&1(;'''C(;1S(&4)&CDD)4C&DCDC444S!('&) F41 ()1C&(4DS(4((&41;;;)(;(DS&D'D4''F&C(DF1;);4(S!('&) 1'' ()4()C)1S(4((&C)&'''4D'4S(F;1'DF()4&'D'D1F)'S!('&) 1)' ()4'C')4S(4((&D'C'''4C)(SF&44DC)'1D1&444(111S!('&) 41C ()4F;)FDS(4((;'(';;)('4&S(;FD1D''C;C&;(FFF(S!('&) C'' ()4''41CS(4((;'1D114);)1S()'FFF;;&'1)4'&DC(&S!('&) ('&) ()4''CD'S(4((;()4;;)'D;FS1'(4&('DCC)&CD)D'4;S!('C' (&'' ()1C&((;S(4((;&'&'''4F)4S(F)(11;F)(;1(1(D(''S!((&' C;& ()4'1C1&S(4((;&F)'''4)DFS&D;)&C(1;CC()D'4F44S!((;1 1)' ()4(((CFS(4((;;11'''4;C;S&1;1;))D1&&&);F;FD)S!((F& 4&' ()4''DD4S(4((;)(&;;)'14'SF'C(;)';C'D&'44DC;'S!((F& 41C ()4F;4F1S(4((;F(('''4&;CS1DD&1(D;1F();4(&D;&S!((F& C1) ()4F;D()S(4((;FD(114;C&)SD(&(D41C'(4)D41C;'DS!((F& D'' ()1;;;C1S(4;C4(D'D4;;FFCS114F4;14())1C&'1&&CS!(&C' 4&' ()4(&)4CS(4;C4&4&;'11C('S(&&(14D;C;(4;'1''DS!(&C' 41C ()1(&)'1S(4;C4;F(D4;;;D4S1('144&C&D1(41F1;D'S!(&C' C'' ()4F;)DFS(4;C4C1(1;DDFF)SCFC)'4'CD1D1;F'44))S!(&C' CF) ()4'14D(S(4;C4D1);'11((CSF&1D1;C)(111CF14'(&S!(&C' D1' ()1C&(;FS(4;CC'&D;'11'F;S));4(C''C''41&CC;'&S!(&C' ('&) ()FCDD44S(4;CC'44;'11''FSD1;;)''44)&CFC'111S!(;;) 4F' ()4(&F(;S(4;CC(D';'1FCD&S(&1DF&';'4&(D)CF1;&S!(;11 41C ()1C&()DS(4;CC;'&1;DD((;SC&DD&;'1C(&;C;F;D'S!()'' ('F' ()4(;'4CS(4)C&;(F1;'F(''S4FD1F'1D'FFF;4&;CC)S!())' D'' ()1C('('S(4)C&);D1;')D41S)CDC4;44D141'4CC)DS!())' D1' ()4(&D&'S(4;CC4'F1;DC4('S)FF(D)&(F(1D1'FCD')S!())' ('&) ()1D(;4DS(4;CC44)1;DC1)(SCDF4'')4)(1CC'))(()S!())' ('C' ()1(F4(4S(4;CCCF1;'1F&&1SF)4;4(D(F4';&'F'(CDS!(1'' 41C ()1'C4C)S(4;CCD''1;DCF(FSC1;;D(C;&(;4((FF1()S!(1'' D'' ()1D()'DS(4;CD'CF;'1)DD4S&4&&C'4D)CF4')4)F;'S!(1'' ('&) ()1;;(4CS(4;CD()C;'1)D;)S)4)4&)CF)1D44)'C'&FS!(1'' (&'' ()4'1D&)S(4;CD&&)D4;(F&)S))41(1D&1);14C;C1FCS!(1'' (&C' ()1(F4D(S(4;CD;(;D4;();FS((CFFF&;4C1CC(C)'DFS!(1C' ('F' ()1D(;1(S(4;CD;CDD4;(;FDS&CF)1D;)1F111)(('(1S!(441 (''' ()1D();4S(4;CD)&CD4;(;&'S11&4D))(;4&C;;;)F;(S!(4D& (;)) ()1C(DD;S(4;CD)4F1;D4D)'S;);'D1F14;4(14F;1C4S!(C'' ())' ()4(('1)S(4;CDF&F1;D4CD'S1'1F()F&D;'C1)4&F''S!(CF1 (;D&

PAGE 42

;; ()1C&(4DS(4;CDF1C1;D4C)4S;4C(F;FD&14444&()4S!(D&' ('C' ()4F;4(CS(4;CD1)1;'1));1SD4F;14&1;14(;44)44S!(D&' (&'' ()4'4C&4S(4;CD4D;;'1)&CDS;C1&&4;);(&1(;)C4FCS!(D&' (&C' ()4((('4S(4;CDCDF1;D4F&'SD(D41'C44CF&'F&C4F1S!(D&' ()'' ()4()C1'S(4;CDD1;1;D4)F&SFD)')4(&C);1;CC;C4)S!(D&' ())' ()F);4';S(4;D'';1;'1)')1S&;(1F)F'F1;&;&CCD41S!&')C ((F& ()4F;&&;S(4;D'C&D;'1;&F;S&&'FD4F(&'&'C)D&DDS!&')C (&C' ()4F;;C(S(4;D'DFF;'1;(&4SF(&))11&D';4C1D&44)S!&')C (F;1 ()4(&F4&S(4;D('4&D4&D141SCF;CF4)D'';DF41C(;(S!&(1' ())' ()1D&('FS(4;D((&C;'1&DF)S(&4)1)(4D1('4)DC&D(S!&;') ())' ()4'4'('S(4;D(;&1;'1&4F1S(1&&))D4');)1;;;;(4S!&;') (4&C ()1D&(&&S(4;D(;F1D4&D;D&S114'(CC4(11C(&D;&1'S!&F;C ('C' ()FCDDF;S(4;D()()D4&D;;)S1DF1&4)'&1)&4)F&1C)S!&F1' ('C' ()4(;'D;S(4;D(F'DD4&D&;DS1)C(&FD1F&')D';F&;)S!&F1' ())' ()4F;;)CS(4;D(F1D;'1&F(;S(F&)D&)'DD4;;4(1&&1S!&F1' (1'' ()1;;&')S(4;D(1(4D4&D(;(S(;)&1D4&D1;4;'1D4)DS!&F1' (4'' ()1C'C14S(4;D(4'(1;DF4()S1)'(&(;D;F;D;&1C11DS!&F1' (C'' ()4F;(D4S(4;D(4F4;'1&;&FS)1F)4);;(1';CF4)(41S!&F1' (D&' ()1;;FDFS(4;D(C'4;'1&&4FS(')D&C11&;1FFD;DC(1S!&F1' &')C ()4(&414S(4;D);(;1;D;('&S&(DF;D;)D'&4F4)FC'&S!&4;& &')C ()1'C41;S(4;D);4D;'FD4';SF((C;C1)111('C4'(FS!&4;1 (C&) ()4'1DC(S(4;D)F1);'FDF(CS))'(F))DD4);;4'(&)'S!&C'' &('' ()1D(D;1S(4;D)1;D;'FD));S;C;4;D11(''&(4C1CFFS!&CC' D'' ()1C&(4(S(4;D)4';D4&1')FS&'F1(DF4(DC'DF(&F')S!&CC' (C'' ()1D(&D(S(4;D)4C)D4&FD1)SFD;D4)(;D'(C1;CD4()S!;''' &''' ()4()DF4S(4;D)CC(;'FD&'(S('CC()D'44'4'&(1&D)S!;&'' (C'' ()4(&D4'S(4;D)D;&;'FD(F'S&1(CC;F()C&1FD4111(S!;&'' &')C ()1C'C1&S(4;D)D4D1;D&);1S&((C&1C'F4)4&;&C'1;S!;&'' &)'' ()4'C''1S(4;DF(1&1;D&&F;S&;D;D((((C1(D;1';'FS!;))' ())' ()1D(D);S(4;DF&1C;'FCC()S;)(FC'(''C;'&FF4141S!;C)' &(1' ()4(&1&4S(4;DF);(1;D(DC)S&14D'1)FC1'&C));)))S!;C)' &)'' ()4'44C(S(4;DF)DC;'FCFC)S1(F(D1&'1F4DD;C;)D;S!)'D1 &;') ()F14)1)S(4;DFFF)1;D(C1(S4&CCC(D;C&FCD'CCC(FS!)'D1 ;'4& ()4(F'&1S(4;DFFD)1;D(C&(S;D1;)4D)D(4()C(1))1S!F(&' &(1' ()4(;'C1S(4;DF1;)1;D(4C(SF4(1414D)D'4F4)(1;(S!F(&' &CC' ()1(F4&4S(4;DF4(;;'FC;1DSF)4(D'4D&CD&;()&C4CS!F(&' ;&'' %.)4%-!!7.%+3 #,$B%%. %,+%$+!! N,

PAGE 43

;) nn #*. "!+ #" #%%., #% $%,.!"%+%$$ */.!%,,!% % +!%%.*+ #!, %%.%%"",*0*3% %,! #"!..!3$ #*/.!%,%,,!3.!%,-! +% !-%!(6+%$ ,!3.!%,, !%+!.,-&6%+!.,-* /.!%,%+%$ #% ,!3.!%,, %,7 !/-!+/* %,;6+%$! #*/.!%,,% ,,!3.!%,,"!+,7 !/ -!+/* 3%!, #% +%$,!3.!%,, !%+!. ,-3%.-!,, 3#.+%$,!3.!%,, !%,7 !/-!+/* 3 r!$.-!,,r!$ ,-% !"+%$ #% %.!%,,%"*.. +%$%, #/!$.-!+ -.%% #+%$+!"!+%.!3>!.* %#$#>!.* !+%$%. /-%.-!,$"!+%..+%$%,%.!%, ,"!+ !/ !! !+/:./:. #*. "!+%-#-#+%%,-% !"#! 3 #+%$3/!-,"! 3$!%!*,-!.,-#%% +%..3$3,!3%,%% .!%,%%.$%, !r!$ %, /-+%$/% #3/!-,%,,!3.!%,,%,7 !/, 7 !/#%%.%$3$%%%, -% %7.$# ..!$ !.!%,*.7%+!. #+%$0*%. *"",.$# # -!,-#+% #-%* #+%$3%*/.!%, ,!%+%..-%, # "!-, !,!3.!%, !%.%$3$,-

PAGE 44

;F /!-$F1,"" +%$,+! 3 %-!-.*,, #% #-* !"" +%$,+!!"%/**/.!%,, !%-!!7" !+%+!.,-())'/:. 3#.+%$*/.!%,,"!+%,7 !/-!+/* &' )C/:. nnn@ !+,%"!"* *%-#3!*., !:%+ #/!. !"+!.%.*+ ,!3.!%,3#%," -!+%%.%. !* #!*$# ,,! #!3%$!" $"-%#!*$# #%-# #!*., .. :%+,"!%., %,+,,, + %,% %%%.#%-#3%-%,!* +%$3 #%/! % % !# %+%-##!*.,/"!+,!.%,-%/+%$ %,, +%+.% %,%/% !9%-!!7%/!/*.%!-% .+,% #%%.! # #% ..!!"/!/.*,%.# #!* .,:/.!,%+.%+%%, %%.9,"!+,,,-! #+%+.% 3 #%-#%, %$%+3%!*$# %-!!7/. !"&'(&L2M# %! %.. # %+* #%..!3"! #"*,%+ %.%. #%%,,!3.!%,+%$

PAGE 45

;1 ("# !!"%-!!7U-#!.!$U# *%,%5,6n ,!((&'(1 "!+# /PP333 #$*%,%-!+P -#!.!$P&''4P?* .P&FP+,%3+,% &%.2r-!+/!?*$$.+%$0*%. %,9U+5,6n !+&C&'(1"!+# /PP333 +-!+P,!-* + %/Y,!-S,R(&&F4;1 ;% $.!% %!<% 5 &'((6!-%.!"+%$!. !-%. 3!7 nrrrrnrr r rrn 5//14DI1C)6# /PP,!!$P('(('DP!&'(( (4 )!+/%"!U%-!!73!!+5,6n ,"!+ # /PP3!!+"-!+P-!+/%>"!P F*+%(>)1()>'4F4>4 S& 1!3.5&'(&!F6%-!!7 ..!%,!* $ "rr #$%&r& /(3!7 4**++%,r#%+* #< .!5&'((6%.9$%-!!7r%$:/% !n%. 'rrr()n*+)) rrrnrr)r$rrrrr 5//1(I4'63!7 # /PP,!!$P('(()FP&'1CC(1&'1CC&; C%!5&'(16 nr),$ rr!rn r !" !.!%,!% D,%"!5,6n ,!+&D&'(1" !+# /PP+,%%% PP,%"! ('!!<%-#.,5(DDF6!.!/ %-.!"!2r+%$-!+/! #$&rn 5)6;4;I;C(# /PP,!!$P('(((4P(&&(4&11 ((!,%@%7%!<+5&''46!.!9 % !/%-%, //.-% +/!G*%. !"2r!.!+%$ .n&&&nrrrrnr 'r 5B!.)/B>;CF>B>;CC6# /PP,!!$P('((' DPr&''4);C'';F (&n*.5(;;;DD(4)C6%-!!7* %$ %+"!](..!n ,!;&'(1 "!+# /PP,%.!!7 +-!+P&'(&P')P'DP"%-!! 7>*> %$%+>"!>(>..!P (;85,6n ,!&;&'(1"!+ # /PP3333$,!$P,!-*+

PAGE 46

;4 ()85,6n ,-+(&'(1"!+ # /PP3333$ !$P,!-*+ (F8*@*Q%$2E$8<8*5& '(16!.!+/!!"2r!,, r#! !!..! n&&&nr'r / 516&1C)I&1D1 # /PP,!!$P('(('DPr&'(1&FF(;11