Nom de la housse

Rfrence

Appareils Compatibles

Dimensions Maximales

Name of the case

Compatible with

Maximum dimensions

Casio Exilim EX- Z110 Nikon S700 Canon Ixus 750 - 800is 54001 Casio Exilim EX- Z10 Casio Exilim EX- Z120 Canon Digital Ixus II's Canon Digital Ixus 60 Canon Ixus 60 - 75 Canon Powershot A Serie Canon Digital Ixus 75 Canon Digital Ixus 70 Canon Powershort A610 Canon Digital Ixus 850is Canon Ixus 850is - 900ti 54004 Nikon Coolpix S510 Canon Digital Ixus 95is Canon Powershot SD 1100 Is Canon Ixus 860is - 960ti 54005 Canon Digital Ixus 990is Canon Digital Ixus 860is Casio Exilim EX-S10 Casio Exilim EX-S600 Casio Exilim EX-S880 Casio Exilim EX-Z8 Casio Exilim EX-Z50 Casio Exilim S et Z Srie 54501 Casio Exilim EX-Z60 Casio Exilim EX-Z100 Casio Exilim EX-Z200 Casio Exilim EX-Z500 Casio Exilim EX-Z750 Casio Exilim EX-Z1200 Fuji FinePix Z1 Fuji FinePix A600 Fuji FinePix F10 Zoom Fujifilm FinePix F - A - Z 54201 Fuji FinePix F11 Zoom Fuji FinePix F20 Zoom Fuji FinePix F30 Zoom Fuji FinePix F31D HP Photosmart R817 HP Photosmart R927 Samsung S1030 Panasonic Lumix DMC-FX520 HP Photosmart 50201 Panasonic Lumix DMC-FX33 Samsung NV24HD
Canon Digital Ixus 40 Canon Digital Ixus 500 Canon Digital Ixus 430 Canon Digital Ixus 750 Canon Digital Ixus 65 Canon Digital Ixus 85is Canon Digital Ixus 100is Canon Powershot S80 Canon Digital Ixus 900ti Sony Cybershot DSC-W55 Canon Digital Ixus 80is Sony Cybershot DSC-W55 Canon Digital Ixus 960ti Casio Exilim EX-S100 Casio Exilim EX-S660D Casio Exilim EX-Z5 Casio Exilim EX-Z9 Casio Exilim EX-Z30 Casio Exilim EX-Z55 Casio Exilim EX-Z70 Casio Exilim EX-Z850 Casio Exilim EX-Z600 Casio Exilim EX-Z1050 Fuji FinePix Z4 Fuji FinePix A400 Zoom Sony Cybershot DSC-W200 Fuji FinePix A470 Zoom Fuji FinePix Z5FD Fuji FinePix A350 Zoom HP Photosmart R818 HP Photosmart R727 HP Photosmart E327 HP Photosmart R417 HP Photosmart R717 Samsung NV30
Canon Digital Ixus 50 Canon Digital Ixus 700 Canon Digital Ixus 30 Canon Digital Ixus 800is Canon Digital Ixus x 59 x 23 mm Canon Powershort A710 Canon Digital Ixus 90is Sony Cybershot DSC-WXx 60 x 26 mm 104 x 66 x 50 mm 90 x 60 x 30 mm
Canon Digital Ixus 970is Casio Exilim EX-S500 Casio Exilim EX-S770D Casio Exilim EX-Z7 Casio Exilim EX-Z11 Casio Exilim EX-Z40 Casio Exilim EX-Z57 Casio Exilim EX-Z80 Casio Exilim EX-Z1000 Casio Exilim EX-Z700 Casio Exilim EX-Z1080 Fuji FinePix Z3 Fuji FinePix A500 Zoom Sony Cybershot DSC-W300 Fuji FinePix A350 Zoom Fuji FinePix Z2 Fuji FinePix A345 Zoom HP Photosmart M527 HP Photosmart R725 HP Photosmart M525 HP Photosmart E317 HP Photosmart M425 Samsung NV40

100 x 60 x 30 mm

96 x 61 x 25 mm

95 x 62 x 35 mm

100 x 65 x 35 mm
Samsung NV15 Panasonic Lumix DMC-FX500 Panasonic Lumix DMC-FX150 Kodak Easyshare V530 Kodak Easyshare V et Pentax Optio W Sries Kodak Easyshare VKodak EasyShare V1233 Kodak EasyShare V1253 Pentax Pentax Optio Wpi Kodak Easyshare C530 Kodak Easyshare C Sries 54602 Kodak Easyshare C875 Kodak Easyshare C533 Kodak EasyShare V1273 Nikon Coolpix S9 Nikon Coolpix 5600 Nikon Coolpix L2 Nikon Coolpix L4 Nikon Coolpix L12 Nikon S - P - L Sries 53801 Nikon Coolpix P4 Nikon Coolpix S3 Nikon Coolpix S7c Nikon Coolpix S600 Nikon Coolpix S710 Nikon Coolpix S550 Nikon S50 Sries 53802 Sony T900 Nikon Coolpix S640 Nikon Coolpix S51C Olympus Mj 1000 Digital Olympus Mj 53901 Olympus Mj 725 Digital Olympus Mj 800 Panasonic Lumix DMC-FX01 Panasonic Lumix DMC-FS2 Panasonic Lumix DMC-FS20 Panasonic FX et FS Sries 51901 Panasonic Lumix DMC-FX07 Panasonic Lumix DMC-FX3 Panasonic Lumix DMC-FX30 Panasonic Lumix DMC-FS6 Panasonic LX Sries 51902 Panasonic Lumix DMC-FT1 Panasonic Lumix DMC-LX1 Panasonic Lumix DMC-TZ2 Panasonic Lumix DMC-TZ3 Panasonic TZ et LZ Sries 51903 Panasonic Lumix DMC-TZ4 Panasonic Lumix DMC-LX2 Panasonic Lumix DMC-TZ6 Panasonic Lumix DMC-TZ5

Pentax Optio A30 Pentax Optio A20 Pentax Optio E20 Pentax Optio T10 Pentax Optio S60 Pentax Optio S45 Pentax Optio M50 Samsung Digimax V700 Samsung Digimax L55W Samsung NV 3 Samsung NV 4 Samsung ST550 Samsung TL225 Samsung i7 Samsung WB210 Samsung NV 11 Sony Cybershot DSC-W115 Sony Cybershot DSC-N1 Sony Cybershot DSC-T77 Sony Cybershot DSC-T300 Sony Cybershot DSC-W125 Sony Cybershot DSC-W130 Sony Cybershot DSC-W170 Sony Cybershot DSC-T90 Sony Cybershot DSC-W200 Sony Cybershot DSC-W90 Sony Cybershot DSC-T90 Sony Cybershot DSC-T10 Nikon P5100 Canon Powershot G10 Canon Powershot G12 Canon Digital Ixus 110 IS Canon Digital Ixus 960ti Nikon Coolpix S710 Panasonic Lumix DMC-FS12 Panasonic Lumix DMC-FS25 Panasonic Lumix DMC-FX550 Samsung ST550 Sony Cybershot DSC-TX1 Sony Cybershot DSC-T90 Sony T900
Pentax Optio A40 Pentax Optio M40 Pentax Optio M20 Pentax Optio S7 Pentax Optio Svi Pentax Optio S55 Pentax Optio S55 Samsung Digimax V800 Samsung Digimax A503 Samsung NV 10 Samsung i8 Samsung TL34HD Samsung ST500 Samsung NV 100 HD Samsung NV 5 Sony Cybershot DSC-W120 Olympus Mj 1030SW Sony Cybershot DSC-T100 Sony Cybershot DSC-T33 Sony Cybershot DSC-T700 Sony Cybershot DSC-W50 Sony Cybershot DSC-W70 Sony Cybershot DSC-W80 Sony Cybershot DSC-W85 Sony DSC-TX10 Sony Cybershot DSC-T900 Nikon Coolpix S200 Nikon P6000 Canon Powershot G7 Canon Digital Ixus 120 is Canon Digital Ixus 970is Nikon Coolpix S50c Nikon Coolpix S630 Nikon S70 Panasonic Lumix DMC-FX50 Samsung TL34HD Sony Cybershot DSC-W300 Samsung IT100 Samsung TLx 60 x 30 mm Nikon P7000 Olympus X-Z1 Canon Powershot S95 Sony Cybershot DSC-TX9 Canon K/Ixus 1000 HS 115 x 80 x 60 mm 95 x 57 x 24 mm 95 x 56,5 x 23,3 mm 100 x 65 x 30 mm Samsung EX1 Sony DSC-HX5V Sony DSC-HX7V Sony DSC-TXx 57 x 31 mm 95 x 65 x 35 mm

Samsung L et I Sries

100 x 62 x 25 mm

Samsung NV Sries

110 x 60 x 65 mm
Copyright 2005-2011 Innovacases / Norve Saint-Tropez - All other products and brand names may be trademarks of their respective owners

Digital Image Authentication from Thumbnails
Eric Kee and Hany Farid Department of Computer Science, Dartmouth College, Hanover NH 03755, USA

ABSTRACT

We describe how to exploit the formation and storage of an embedded image thumbnail for image authentication. The creation of a thumbnail is modeled with a series of ltering operations, contrast adjustment, and compression. We automatically estimate these model parameters and show that these parameters dier signicantly between camera manufacturers and photo-editing software. We also describe how this signature can be combined with encoding information from the underlying full resolution image to further rene the signatures distinctiveness. Keywords: Digital Image Forensics, Digital Image Authentication

1. INTRODUCTION

Many organizations are struggling with the issue of photo tampering. For example, digital images, videos, and audio are now routinely introduced as evidence in civil, criminal, and national security cases. In such cases, the integrity of digital evidence is central. Digital forensic techniques have been developed to detect: region duplication1, 2 ; inconsistencies in camera response function3 ; inconsistencies in lighting4 ; and inconsistencies in sensor noise5 see6 for a general survey. However, relatively benign modications such as cropping or contrast adjustment either cannot be detected by these techniques, or render these techniques ineective. In the legal setting we are sometimes interested in determining if a digital image has been altered in any way from the time of recording, including something as simple as cropping. Here we approach this problem of image authentication by exploiting a cameras customized image encoding format. We describe how to exploit the formation and storage of an embedded image thumbnail for image authentication. Specically, we model the creation of a thumbnail with a series of ltering operations, contrast adjustment, and compression. We describe a technique for automatically estimating these model parameters, and show that these parameters, although not unique, can dier signicantly between camera manufacturers and photo-editing software, Figure 1. As such, these parameters can be used as a signature for image authentication. We also describe how this signature can be combined with encoding information from the underlying full resolution image7 to further rene the signatures distinctiveness.

2. METHODS

Embedded in an image header is a thumbnail version of the full resolution digital image. We describe a model for the creation of a thumbnail, and then describe how to estimate these model parameters.

2.1. Thumbnail Model

A thumbnail is typically on the order of pixels in resolution. Given an image f (x, y), the thumbnail is created by a series of six steps: crop, pre-lter, down-sample, post-lter, contrast and brightness adjustment, and JPEG compression. If the original resolution image is of a dierent aspect ratio than the thumbnail, then the image needs to either be padded or cropped accordingly. The amount of padding/cropping is specied by four parameters cl , cr , ct , cb , where cl and cr correspond to the padding/cropping on the left and right, and ct and cb on the top and bottom, Figure 2. A positive value corresponds to a padding (with a pixel value of 0), and a negative value corresponds to a cropping. Denoting the cropped image as f (x, y), the next four processing steps are specied as follows: t(x, y) = D{f (x, y) h1 (x, y)} h2 (x, y) + , (1)
Contact: kee@cs.dartmouth.edu and farid@cs.dartmouth.edu
Figure 1. Shown from left to right are an original thumbnail generated by a Nikon D200, a thumbnail of the same image generated by Adobe Photoshop CS3, and the dierence between these thumbnails. The Nikon thumbnail is pixels in size, while the Photoshop thumbnail is 160 107. In addition these JPEG encoded thumbnails employ dierent cropping boundaries and quantization tables leading to further dierences as can be seen in the right-most panel.
where t(x, y) is the thumbnail, h1 () is the pre-lter, D{} is the down-sampling operator, h2 () is the post-lter, is the convolution operator, and and are the multiplicative contrast and additive brightness adjustment terms, respectively. The pre-lter is typically a low-pass lter applied to avoid spatial aliasing prior to downsampling, and the optional post-lter is typically a sharpening lter. In the nal step, the thumbnail is JPEG compressed with a specied quantization table. In order to simplify this model, a series of assumptions are made: the pre-lter is assumed to be a circularly symmetric Gaussian, exp((x2 + y 2 )/ 2 ), with width the pre-lter is unit sum the post-lter is pixels in size the post-lter is symmetric (h2 (x, y) = h2 (x, y) and h2 (x, y) = h2 (x, y)), yielding a lter of the form (a b a ; b c b ; a b a) the post-lter is unit-sum, constraining c = 1 (4a + 4b).

With these constraints, the full model for creating a thumbnail is specied by 11 processing parameters: 2 for the size of the thumbnail, 4 for the cropping/padding, 1 for the pre-lter, 2 for the post-lter, 2 for the contrast and brightness. In addition, there are 128 compression parameters: the JPEG quantization table is specied by two tables corresponding to the quantization for the luminance and chrominance channels. This yields a total of 139 model parameters. In the following section we describe how to estimate these model parameters.
2.2. Thumbnail Estimation
In the rst step of thumbnail construction, a rectangular cropping boundary relative to the full resolution image is specied. This cropping boundary is determined by anisotropically scaling and translating a bounding box of the same size as the full resolution image such that the cropped and downsampled image matches the extracted thumbnail, Figure 2. The cropping parameters are estimated by rst specifying an initial boundary that has the same aspect ratio as the thumbnail, is scaled to encompass the maximum image dimension, and is translated such that the image is centered within the boundary, Figure 2. The full resolution image f (x, y) is then cropped according to this boundary, padded with zeros (where necessary), and downsampled to yield an initial thumbnail: t0 (x, y) = D{f0 (x, y)}, (2)
where f0 (x, y) is the initial cropped image, and the downsampling rate is max(Nx /nx , Ny /ny ), where (Nx , Ny ) and (nx , ny ) are the image and thumbnail dimensions, respectively. This initial thumbnail is then anisotropically scaled and translated to match the extracted thumbnail t(x, y): t(x, y) =
t0 (sx x + x , sy y + y ),
We assume that the two chrominance channels employ the same quantization table.
Figure 2. Creating the crop boundary for the full resolution image f (x, y). Top row: create an initial crop boundary f0 (x, y) so that its aspect ratio matches the thumbnail t(x, y); down-sample to create t0 (x, y); align with the actual (x, y). Bottom row: adjust the initial crop thumbnail t(x, y) by scaling and translating by M = (sx , sy , x , y ) to yield t boundary f0 (x, y) by M to yield the desired crop boundary f (x, y). The crop boundary is specied by the margins cl , cr , ct , cb.
where (sx , sy ) and (x , y ) are the scaling and translation parameters. These parameters are estimated using a coarse-to-ne dierential registration technique.8, 9 This registration is performed on a grayscale version of the color thumbnail. The scaling and translation parameters are then used to specify the cropping boundary for the full resolution image. Specically, the coordinates of the four corners of the initial cropped image f0 (x, y) are scaled and translated by (sx , sy ) and (x , y ), yielding the desired rectangular cropping boundary f (x, y), Figure 2. This boundary is parametrized by the horizontal and vertical margins cl , cr , ct , and cb , specied as a percentage of the image dimensions, Figure 2. Note that when down-sampling f (x, y), the sampling rate in the horizontal and vertical directions must be independently adjusted so that the nal dimensions match that of the thumbnail t(x, y). Next, the pre-lter h1 (), post-lter h2 (), and contrast and brightness terms and are estimated by minimizing the following error function: E1 (h1 , h2 ) =

t(x, y) D{f (x, y) h1 (x, y)} h2 (x, y)
Because the translation parameters (x , y ) are specied in thumbnail coordinates, they must be scaled by the downsampling rate between f0 (x, y) and t0 (x, y).
Note that this error function is only specied in terms of the pre- and post-lter, and not the contrast and brightness. Given a pair of lters, the contrast and brightness are estimated by minimizing the following error function: E2 (, ) =

t(x, y) (th (x, y) + )

where th (x, y) = D{f (x, y) h1 (x, y)} h2 (x, y). This error function is minimized using standard least-squares estimation. The summation in E1 () and E2 () are performed over all three color channels of the thumbnail. The error function E1 () is minimized using a brute-force search over the lter parameters, where on each iteration of this search, the error function E2 () is minimized analytically to yield the best contrast and brightness terms for the specied lters. These parameters are rened using an iterative Nelder-Mead minimization, which is bootstrapped with the results of the brute-force minimization. In practice, we have found that the minimization of E1 () is slightly more eective when performed in the Fourier domain. Specically, we minimize: E1 (h1 , h2 ) =
W (x , y ) |T (x , y )| |T (x , y )|
where T () is the Fourier transform of the actual thumbnail t(), T () is the Fourier transform of our constructed h1 } h2 ) + , and || denotes the magnitude of the Fourier transform. This error function thumbnail (D{f is frequency weighted with a highpass lter, W (), because after the initial alignment of Equation (3) the lowfrequencies are already well aligned. Lastly, the thumbnail dimensions and 128 thumbnail compression parameters are extracted directly from the JPEG header. Note that since the original thumbnail t() was compressed with these parameters, we must also compress our estimates of the thumbnail during the parameter estimation stages described above. Specically, t0 () is compressed prior to estimating the scaling and translation parameters, and th () is compressed prior to estimating the contrast and brightness terms.

3. RESULTS

To verify that thumbnail parameters can be reliably used for image authentication, a database of 1514 unmodied images were downloaded from Flickr. An image was considered to be unmodied if the image was tagged by Flickr as original, no metadata elds had been removed, the metadata modication and original dates were consistent, and the metadata software eld was empty. These 1514 images spanned 142 cameras of dierent make and model, Appendix A. Cameras of the same make and model sometimes vary their image and thumbnail size and quantization table. Because these variations aect the overall thumbnail parametrization, we partitioned the 1514 images into 245 camera classes containing images from the same make and model and image/thumbnail size and quantization table. For computational eciency the thumbnail parameters were rst estimated for a representative image from each of the 245 camera classes. The crop boundary was estimated as described in the previous section. The pre-lter, post-lter, and contrast and brightness were estimated using a brute-force search, Equation (4). The parameter space for the pre-lter was partitioned into the range [0.005, 1] in 20 equal steps. The parameter space for the two components of the post-lter were partitioned into the range [0.5, 0.5] in 11 equal steps. Recall that the contrast and brightness terms were estimated analytically within each iteration of the brute force search, Equation (5). These brute force solutions were then used to bootstrap the Nelder-Mead minimization for the remaining images within each camera class.

Variations in image/thumbnail size and quantization table can, for example, be due to dierences in rmware. In order to appropriately compare the width of the pre-lter h1 () across images of varying size, the size of the pre-lter h1 () is normalized relative to the image dimensions. Specically, the pre-lter is sampled over the interval [1, 1] at a rate 1/150 of the maximum dimension of the original image.

0.8 Probability

40 Equivalence Class Size
Figure 3. Shown in the left column is the probability that a camera is contained in an equivalence class of a specied size. Shown in the right column is the number of equivalence classes of a specied size. Shown along the top row are the distributions based on all of the thumbnail parameters (40.8% of the cameras have unique parameters, i.e., an equivalence class size of one). Shown in the bottom row are the distributions for only the thumbnail size and quantization table (23.7% of the cameras have unique parameters).
Because the absolute value of the contrast and brightness terms may depend on the underlying image content, these parameters are combined into a single binary-valued parameter corresponding to the presence or absence of contrast/brightness adjustment. Specically, a camera is said to apply a contrast/brightness adjustment if the cameras contrast deviates by more than 0.075 from unity, or if the brightness deviates by more than 0.05 from zero (assuming a luminance scale of [0, 1]). These thresholds were determined by rst computing the dierence between each images thumbnail parameter and the mean parameter of the corresponding camera class. The threshold was then taken to be the width of the 99% condence interval of the distribution of these dierences. We next evaluate the consistency and distinctness of the thumbnail parameters within and across camera classes. The thumbnail parameters for a given camera class are determined by computing the mean of the crop boundary, the mean of the pre- and post-lters, and the mode of the contrast/brightness, across all images in a camera class. The thumbnail size and quantization table are, by denition, the same for all images in a camera class. In order to reasonably equate real-valued model parameters (crop boundary, pre-lter, post-lter), these parameters were considered as equivalent if their absolute dierence was below a specied threshold. The
0.8 Probability Probability 5 Equivalence Class Size 6

5 Equivalence Class Size

Figure 4. Shown on the left is the probability that a camera is contained in an equivalence class of a specied size based on all of the image and thumbnail parameters (72.2% of the cameras have unique parameters, i.e., an equivalence class size of one). Shown on the right is the cumulative probability distribution.
threshold for the crop boundary was 0.01, the pre-lter 0.2, and the post-lter 0.075. These thresholds were determined in the same way as for the contrast and brightness thresholds described above. Two thumbnail models were considered to be equivalent if their integer-valued parameters (contrast/brightness, size, quantization table) were the same and if their real-valued parameters (crop boundary, pre-lter, post-lter) were within the specied thresholds described above. Cameras were then grouped into equivalence classes of identical models. Shown in the top left panel of Figure 3 is the probability that a camera is contained in an equivalence class of a specied size based on using all of the thumbnail parameters: 40.8% of the cameras are in an equivalence class of size one (i.e., are unique), 9.8% are in an equivalence class of size two, 8.2% are in an equivalence class of size three, and the largest equivalence class is of size 48, with 19.6% of the cameras. There is only one class of size 48 and it contains 42 Canon Powershots and 6 Canon Digital Ixus of varying models. Shown in the top right panel of Figure 3 is the distribution of equivalence class sizes (i.e., the number of equivalence classes of each size). For comparison, shown in the bottom row of Figure 3 are the distributions for only the thumbnail size and quantization table: 23.7% of the cameras are in an equivalence class of size one, 8.25% are in an equivalence class of size two, 7.4% are in an equivalence class of size three, and the largest equivalence class is of size 56, with 22.9% of the cameras. Note that the addition of the thumbnail processing parameters improves the distinctiveness of the signature. Shown in Figure 4 is the probability that a camera is contained in an equivalence class of a specied size based on using all of the thumbnail parameters and the full resolution image size and quantization table. In this case, 72.2% of the cameras are in an equivalence class of size one (i.e., are unique), 11.4% are in an equivalence class of size two, 6.1% are in an equivalence class of size three, and the largest equivalence class is of size 6, with 2.5% of the cameras. The addition of the image parameters signicantly improves the distinctiveness of the signature. A closer look at these equivalence classes reveals some interesting trends. There is one equivalence class of size 6 and it contains six 8-megapixel Canon Powershot cameras of varying models. There are three equivalence classes of size 5: one contains 8-megapixel Canon Powershot and Canon Ixus cameras and the remaining classes each contain a mixture of 7-megapixel Canon Powershot and Canon Ixus cameras of varying models. With only two exceptions, the remaining equivalence classes of size greater than 1 are populated with Canon cameras which seem to be particularly consistent in their thumbnail and image parameters. See Appendix A for a complete break-down of these distributions.

Figure 5. Shown on the left is the probability that a camera is contained in an equivalence class of a specied size based on only the image size and quantization table (40.4% of the cameras have unique parameters, i.e., an equivalence class size of one). Shown on the right is the cumulative probability distribution.
Lastly, shown in Figure 5 is the probability distributions based on using only the full resolution image size and quantization table. In this case, 40.4% of the cameras are in an equivalence class of size one, 18.8% are in an equivalence class of size two, 18.4% are in an equivalence class of size three, and the largest equivalence class is of size 6, with 9.8% of the cameras. Note that the addition of the thumbnail parameters, Figure 4, signicantly improves the distinctiveness of the signature. To further test the utility of thumbnail parameters for authentication, we analyzed the thumbnail parameters used by Photoshop (CS3). An image was saved with Photoshop at each of 13 possible JPEG quality settings. The thumbnail and image parameters were estimated for each image and compared to the remaining 245 camera classes. None of the Photoshop thumbnail parameters were shared by any of the camera classes. We also individually compared the Photoshop parameters to the camera classes: None shared the same image quantization table, thumbnail quantization table, or pre/post lter; eleven of the camera classes used the same crop parameters: nikon d300 and coolpix l4; fujilm nepix s5700/s700, s100fs, and z20fd; canon powershot sd400 and s2 is; casio ex-z77, ex-z1050, ex-s880, and ex-z60; the olympus e-510 and the sony dsc-w120.

4. DISCUSSION

An image thumbnail is created by a series of six steps: crop, pre-lter, down-sample, post-lter, contrast and brightness adjustment, and JPEG compression. We have described how these parameters can be estimated, and have shown that they vary signicantly between camera manufacturers and photo-editing software. The distinctiveness of these parameters becomes even more pronounced when they are combined with the full resolution image size and compression parameters. As such, these thumbnail and image parameters can be used for image authentication (determining if an image was altered in any way from the time of its recording). The eectiveness of this approach depends on building a library of thumbnail and image parameters extracted from a large collection of images. We have collected 1.2 million original images from which we plan to build this library. From our initial tests, we have found that 83% of the images have a thumbnail. For the images without thumbnails, the full resolution image parameters can still be used for authentication.

ACKNOWLEDGMENTS

This work was supported by a gift from Adobe Systems, Inc., a gift from Microsoft, Inc., and a grant from the National Science Foundation (CNS-0708209).

REFERENCES

1. J. Fridrich, D. Soukal, and J. Luk, Detection of copy move forgery in digital images, in Proceedings of as Digital Forensic Research Workshop, August 2003. 2. A. Popescu and H. Farid, Exposing digital forgeries by detecting duplicated image regions, Tech. Rep. TR2004-515, Department of Computer Science, Dartmouth College, 2004. 3. Z. Lin, R. Wang, X. Tang, and H.-V. Shum, Detecting doctored images using camera response normality and consistency, in Computer Vision and Pattern Recognition, (San Diego, CA), 2005. 4. M. Johnson and H. Farid, Exposing digital forgeries in complex lighting environments, IEEE Transactions on Information Forensics and Security 3(2), pp. 450461, 2007. 5. J. Luk, J. Fridrich, and M. Goljan, Digital camera identication from sensor noise, IEEE Transactions as on Information Security and Forensics 1(2), pp. 205214, 2006. 6. H. Farid, A survey of image forgery detection, IEEE Signal Processing Magazine 2(26), pp. 1625, 2009. 7. H. Farid, Digital image ballistics from JPEG quantization: A followup study, Tech. Rep. TR2008-638, Department of Computer Science, Dartmouth College, 2008. 8. E. Simoncelli, Handbook of Computer Vision and Applications, ch. Bayesian Multi-scale Dierential Optical Flow, pp. 397420. Academic Press, 1999. 9. H. Farid and J. Woodward, Video stabilization and enhancement, Tech. Rep. TR2007-605, Department of Computer Science, Dartmouth College, 2007.

Appendix A

Each entry in this table corresponds to an equivalence class of identical thumbnail and image parameters, Figure 4. The size of the equivalence class is in the rst column and the make, model, and resolution are in the second column. For compactness, all cameras in an equivalence class of size one are listed in a single entry for these cameras, the parenthetical following the make and model corresponds to the number of instances of the same make and model with dierent thumbnail and image parameters. canon powershot sd870 is (8mp) canon powershot sd850 is (8mp) canon powershot s80 (8mp) canon powershot a590 is (8mp) canon powershot sx100 is (8mp) canon powershot a720 is (8mp) canon powershot sd1100 is (8mp) canon powershot s5 is (8mp) canon powershot a630 (8mp) canon powershot a580 (8mp) canon digital ixus 80 is (8mp) canon powershot sd750 (7mp) canon powershot sd800 is (7mp) canon powershot a570 is (7mp) canon powershot sd550 (7mp) canon digital ixus 850 is (7mp) canon powershot a710 is (7mp) canon digital ixus 70 (7mp) canon powershot a620 (7mp) canon powershot tx1 (7mp) canon powershot a550 (7mp) canon powershot a530 (5mp) canon powershot a610 (5mp) canon powershot sd400 (5mp) canon powershot a460 (5mp) canon powershot sd600 (6mp) canon powershot a540 (6mp) canon powershot sd630 (6mp) canon eos digital rebel xti (10.1mp) canon eos 400d digital (10.1mp) canon eos kiss digital x (10.1mp) canon eos digital rebel(6.3mp) canon eos 300d digital (6.3mp) canon eos 10d (6.3mp) sony dsc-w55 (7.2mp) sony dsc-h5 (7.2mp) sony dsc-w35 (7.2mp) canon powershot s3 is (6mp) canon digital ixus 800 is (6mp) canon digital ixus 60 (6mp) canon powershot sd600 (6mp) canon powershot s3 is (6mp) canon eos digital rebel xt (8mp) canon eos 350d digital (8mp) canon eos 30d (8.2mp) canon eos 20d (8.2mp) canon eos digital rebel xsi (12.2mp) canon eos 450d (12.2mp) canon eos digital rebel xsi (12.2mp) canon eos 450d (12.2mp) canon powershot a510 (3.2mp) canon powershot a95 (5mp) canon powershot sd450 (5mp) canon powershot s2 is (5mp) canon powershot g9 (12.1mp) canon powershot sd950 (12.1mp) canon powershot a520 (4mp) canon powershot a85 (4mp) canon powershot a75 (3.2mp) canon powershot sd200 (3.2mp) canon powershot sd1000 (7.1mp) canon powershot s230 (3.1mp) canon powershot a640 (8mp) canon powershot sd790 is (10.1mp) kodak easyshare m1033 (10mp) kodak easyshare z1012 (10mp) canon powershot g7 (10mp) canon powershot sd900 (10mp)