CN1916958A - Method of detecting distorts of JPEG image - Google Patents
Method of detecting distorts of JPEG image Download PDFInfo
- Publication number
- CN1916958A CN1916958A CN 200610036602 CN200610036602A CN1916958A CN 1916958 A CN1916958 A CN 1916958A CN 200610036602 CN200610036602 CN 200610036602 CN 200610036602 A CN200610036602 A CN 200610036602A CN 1916958 A CN1916958 A CN 1916958A
- Authority
- CN
- China
- Prior art keywords
- image
- jpeg
- block
- quantization table
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000013139 quantization Methods 0.000 claims abstract description 56
- 238000012360 testing method Methods 0.000 claims description 39
- 238000007906 compression Methods 0.000 claims description 25
- 230000006835 compression Effects 0.000 claims description 23
- 238000001514 detection method Methods 0.000 claims description 10
- 239000004744 fabric Substances 0.000 claims description 9
- 239000003086 colorant Substances 0.000 claims description 5
- 238000011002 quantification Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 8
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 241001270131 Agaricus moelleri Species 0.000 description 1
- 241000272201 Columbiformes Species 0.000 description 1
- 241001504519 Papio ursinus Species 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010008 shearing Methods 0.000 description 1
- 239000007858 starting material Substances 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Images
Landscapes
- Compression Of Band Width Or Redundancy In Fax (AREA)
Abstract
A method for detecting distortion of JPEG image includes confirming original JPEG block boundary position of image, estimating quantization table of JPEG small image block, identifying distorted image and positioning distorted region.
Description
Technical field
The invention belongs to the multi-media information security technical field, be specifically related to a kind of altering detecting method of jpeg image.
Background technology
High-resolution digital camera, popularizing and using of powerful PC and various image processing softwares makes that general user can both realize distorting of digital picture do not stayed obvious marks at an easy rate.Nowadays, soon be not real, if false photograph is abused, will bring some problems, as relate to the authenticity of legal argument, the copyright of Digital Media, individual's secret protection etc., the authenticity of inspection image data has crucial meaning.
JPEG is a Standard of image compression that the scope of application is extensive, and tampered image generally can be processed modification as " starting material " with jpeg image, and it is very necessary to detect distorting of this form.Nowadays, at " blind Detecting " (no watermark and signing messages) technology appearance as yet of jpeg image, and also not high to the algorithm for estimating precision of the little image block quantization table of JPEG.
Summary of the invention
The altering detecting method that the purpose of this invention is to provide a kind of jpeg image can be used for given digital picture, differentiates to obtain after whether it is distorted by jpeg image.
Technical scheme of the present invention is as follows: the inventive method may further comprise the steps: the original jpeg block boundary position of 1) determining testing image; 2) estimate the little image block quantization table of JPEG; 3) identification tampered image and positioning tampering zone.
Step 1) determines that the concrete grammar of original jpeg block boundary position of testing image is as follows: 1. determine the region of search of piece, for its region of search of gray level image 8 * 8, and for the coloured image region of search 16 * 16; 2. determine the size of test block, in order to improve the accuracy of detection, the general selection is not less than the piece of 64 * 64 sizes as test block; 3. determine jpeg block boundary, as initial point, shear size and be 64 * 64 image block with each point of region of search as test block, for (the i of each point in the region of search, j) with the quantization table set A in each quantization table, utilize following formula to determine the initial point of its block boundary locations:
(x,y)=arg(max(C(im
(i,j)(k),im
(i,j))))
Im wherein
(i, j)Be the test pattern piece that does not have once more the JPEG compression, im
(i, j)(k) k quantization table during expression is gathered with quantization table is to im
(i, j)Do the image block that obtains after the JPEG compression, wherein | A|=S, k=1,2...S, C (J
1, J
2) two image block J of expression
1, J
2The number that equates gray-scale value makes the i that following formula satisfies, and (x y) is the starting point coordinate of detected original jpeg block boundary to the j value.
Step 2) specific practice of the little image block quantization table of estimation JPEG is at first testing image to be sheared the capable y row of x, then the image that obtains is carried out piecemeal, and the branch block size is B * B, for each image block J
i, estimate its quantization table, method of estimation is as follows: establishing jpeg image piece to be detected is J
i, the span of its quantization table is in set A, at first image block J
iCarry out the JPEG compression with the quantization table in the set A, obtain S width of cloth image, relatively this S width of cloth image and J
iThe number of equal gray-scale value, select therefrom to equate the maximum image block I of gray-scale value that the quantization table of I is then as image block J
iThe estimation of quantization table.
Step 3) identification tampered image identifies by step 2 for utilizing different colours with the concrete grammar in positioning tampering zone) in detected different quantification table block, if it is inconsistent the color of large area region and background (the maximum piece of quantization table occurring) to occur, then think the jpeg image of being distorted.
Utilize this patent can realize effectively the quantization table of fritter jpeg image is done estimation, can detect the original JPEG border that is sheared image effectively, do not do lossy compression method (as only being converted to harmless picture format) at tampered image, or under the situation of compression little (as quality factor greater than 96), the location of detecting with the tampered region of distorting that can realize jpeg image more effectively.This distinguishes etc. that to relating to the jpeg image true or false problem has important practical sense and practical value are very arranged.
Description of drawings
Fig. 1 is the procedure chart of distorting of jpeg image; Wherein Fig. 1 (a) is original jpeg image, and Fig. 1 (b) is the modification to image, and Fig. 1 (c) is the tampered image through aftertreatment;
Fig. 2 is the synoptic diagram on the original JPEG of search border;
Fig. 3 is the general flow figure that the JPEG compression is handled;
Fig. 4 is the process flow diagram that image is carried out twice JPEG compression;
Fig. 5 is a gray level image quantizing factor drawing for estimate; Fig. 5 (b) expression is estimated diagram to uncompressed image, and Fig. 5 (c) expression is that 75 images are estimated diagram to quality factor.
Fig. 6 is a coloured image quantizing factor drawing for estimate; Fig. 6 (b) expression is estimated diagram to uncompressed image, and Fig. 6 (c) expression is that 90 images are estimated diagram to quality factor.
Fig. 7 is the accuracy average diagram that the image block of different sizes is estimated; Fig. 7 (a) estimates that to the gray level image piece Fig. 7 (b) estimates the coloured image piece;
Fig. 8 is a test example 1; Wherein Fig. 8 (a) is that original image is that quality factor is 70 jpeg image, and Fig. 8 (b) makes local the modification to image, removes the beard of people among the figure, is 100 compression subsequently with quality factor, and Fig. 8 (c) is a testing result;
Fig. 9 is a test example 2; Wherein Fig. 9 (a) is that original image is that quality factor is 30 jpeg image, Fig. 9 (b) is that the Region-Duplication mode of distorting illustrates: utilize a certain fritter of image to duplicate and paste in the image-region of desire removal, wherein Fig. 9 (c) removes the pigeon of image, be 99 to carry out the image that JPEG compression obtains with quality factor subsequently, wherein Fig. 9 (d) is the result that image (c) is detected;
Figure 10 is the synthesis model of distorting of general jpeg image;
Figure 11 is the synthetic detection example of distorting; Wherein Figure 11 (a) and Figure 11 (b) are original images, and Figure 11 (c) is a synthetic back jpeg image, Figure 11 (d) testing result.
Embodiment
Below in conjunction with accompanying drawing the inventive method is described further.
The inventive method may further comprise the steps: the original jpeg block boundary position of 1) determining testing image; 2) estimate the little image block quantization table of JPEG; 3) identification tampered image and positioning tampering zone.
1) determines the original jpeg block boundary position of testing image.As shown in accompanying drawing 1 (c), tampered image might pass through post-processing operation such as shearing, image format conversion, just original jpeg image block boundary and quantization table information can be lost.We will determine at first whether given image is obtained by the JPEG modification, and original block boundary locations where.The method that detects is shown in Figure of description 2.
1. determine the region of search (Searching Range) of piece.For its region of search of gray level image 8 * 8, and for coloured image and since be subjected to down-sampling etc. factor affecting, the region of search should expand 16 * 16 to.
2. determine the size (Testing Block) of test block.Obtain from experimental data, the size of test block is relevant with the accuracy of detection, and test block is big more, and accuracy is high more, otherwise low more.This patent suggestion test block should be not less than 64 * 64.
3. determine jpeg block boundary.As initial point, shear size and be 64 * 64 image block with each point of region of search as test block.For each point in the region of search (i, j) with the quantization table set A in each quantization table.Utilize following formula to determine the initial point of its block boundary locations:
(x,y)=arg(max(C(im
(i,j)(k),im
(i,j))))
Im wherein
(i, j)Be the test pattern piece that does not have once more the JPEG compression, im
(i, j)(k) k quantization table was to im during expression was gathered with quantization table
(i, j)Do the image block that obtains after the JPEG compression, function C (J
1, J
2) two image block J of expression
1, J
2The number that equates gray-scale value.The i that makes the following formula establishment, (x y) is the starting point coordinate of detected original jpeg block boundary to the j value.
2) estimation of quantization table.At first testing image is sheared the capable y row of x, then the image that obtains is carried out zero lap ground piecemeal, the branch block size is B * B.B is more little, and the precision of expression tamper detection piece is high more, otherwise low more.But can know that from experiment B is more little, the accuracy of detection is low more, otherwise high more.For its accuracy of detection of balance and accuracy, simultaneously for consistent with JPEG piecemeal rule, the B requirement is 16 multiple.Therefore this patent is advised gray level image B=16, for coloured image, and B=32.For each image block J
i, utilize following method to estimate its quantization table: establishing jpeg image piece to be detected is J
i, the span of its quantization table is in set A.At first image block J
iCarry out the JPEG compression with each quantization table in the set A, obtain S width of cloth image, | A|=S.Compare this S width of cloth image and J
iThe number of equal gray-scale value, therefrom select gray-scale value to equate maximum image block I, the quantization table of I is then as image block J
iThe estimation of quantization table.
3) identification tampered image and positioning tampering zone.According to 2) in the estimated value of quantization table, utilize different colours to identify different quantification table blocks, if the color that large area region and background (the maximum piece of quantization table occurring) occur is inconsistent, then think the image that testing image was distorted, the piece of large tracts of land different colours is the zone of distorting.
Fig. 3 is general JPEG compression process, and from its compression process, image is broken down into equal-sized image block, operation independently between piece and the piece.Thereby the intrinsic blocking effect characteristics of formation jpeg image.
Fig. 4 has described the flow process of image being carried out secondary JPEG compression, and we have therefrom proved following inequality:
P(Y=Y′|q
2≠q
1,q
2>1)≤P(Y=Y′|q
2≠q
1,q
2=1)≤P(Y=Y′|q
2=q
1)
Wherein, Y '=d
1' (i, j), Y =d
2' (i, j), q
1, q
2The quantization step of corresponding respectively two JPEG compression, when this inequality is illustrated in JPEG compression for the second time, if the quantization table that uses with compression for the first time is identical, the probability maximum that equates of the DCT coefficient after both corresponding quantitative then, the quantization table (all quantization steps are 1) that takes second place and be to use quality factor 100.
Obtain the criterion of energy estimated image piece quantization table:
C(J
1,J
2)=|E|,E={(x,y)|J
1(x,y)=J
2(x,y),1≤x≤M,1≤y≤N}
J wherein
1, J
2Be two jpeg image pieces, C (J
1, J
2) represent the number that both corresponding plain picture point gray-scale values equate.Both are similar more for the big more expression of its numerical value, so we can utilize known J
2Quantization table estimate J
1The quantization table of image block.
The estimating experiment of quantization table:
Fig. 5 utilizes the estimation diagram of above-mentioned criterion to the gray scale Lena figure quantization table of 256 * 256 sizes, Fig. 5 (a) gray scale Lena image, and the scope value of supposing its quantizing factor is between [1,100].Fig. 5 (b) is the situation to no JPEG compression Lena image detection, and can carry out number that image that secondary JPEG compression obtains equates gray scale to it increases with the increase of quality factor, and its curve is the process of a monotone variation substantially.Fig. 5 (c) is to being the testing result of 75 compressed images with quality factor, can see that obvious variation can appear in curve, and it is on 75 the position that maximal value can appear at corresponding quality factor.Fig. 6 (a) is colored baboon image, and the scope value of establishing its quantizing factor equally is between [1,100].Fig. 6 (b) is the situation that no jpeg compressed image is detected, and Fig. 6 (c) is the testing result to coloured image, and is similar with Fig. 5 testing result.Therefore utilize this patent method can realize estimation to gray scale and colored jpeg image piece quantization table.
For the validity of further test this method and to the detection of little image block, Fig. 7 has provided the image blocks of different sizes has been estimated.Illustrating each curve is the mean value of 100 width of cloth images accuracy under different quantization tables, and Fig. 7 (a) is to gray level image piece testing result, and Fig. 7 (b) is to coloured image piece testing result.Can see from data in graph form and to utilize this patent method can estimate the little image block of JPEG more exactly, when block size greater than 64 * 64 the time, accuracy all is higher than 95%.
The jpeg block boundary test experience:
Utilize the detection of method that the present invention proposes to jpeg block boundary, test as follows: 200 width of cloth images (100 gray scales, 100 colours) compress (quality factor 10-95 with different quantization tables, JPEG Cs6-Cs10 quantization table among the Photoshop, with a self-defining quantization table (all quantization steps are 10), image is sheared randomly then.Following table 1 has provided the accuracy under gray scale, colored each 100 width of cloth image.
Table 1
10 | 30 | 50 | 70 | 90 | 95 | CS6 | CS8 | CS10 | Custom | lossless | |
| 100% | 100% | 100% | 100% | 100% | 100% | 100% | 100% | 100% | 100% | 99% |
Colored | 100% | 100% | 100% | 100% | 100% | 100% | 100% | 100% | 100% | 100% | 100% |
Fig. 8, Fig. 9, Figure 10 and Figure 11 are the testing results of different test examples.Can see at tampered image and not doing under the situation of bigger compression, can obtain the quite good detecting result.
Claims (4)
1, a kind of altering detecting method of jpeg image is characterized in that the inventive method may further comprise the steps: the original jpeg block boundary position of 1) determining testing image; 2) estimate the little image block quantization table of JPEG; 3) identification tampered image and positioning tampering zone.
2, according to the altering detecting method of claims 1 described a kind of jpeg image, it is characterized in that described step 1) determines that the concrete grammar of original jpeg block boundary position of testing image is as follows: the region of search of 1. determining piece, for 8 * 8 zones of its region of search of gray level image in the image upper left corner, and for the coloured image region of search 16 * 16; 2. determine the size of test block, in order to improve the accuracy of detection, the general selection is not less than the piece of 64 * 64 sizes as test block; 3. determine jpeg block boundary, (i j) as initial point, shears size and is 64 * 64 image block as test block im with each point in the region of search
(i, j), for each point in the region of search and each quantization table in the quantization table set A, utilize following formula to determine the initial point of its block boundary locations:
(x,y)=arg(max(C(im
(i,j)(k),im
(i,j))))
Im wherein
(i, j)Be not through the test block of JPEG compression once more, im
(i, j)(k) k quantization table during expression is gathered with quantization table is to im
(i, j)Do the image block that obtains after the JPEG compression, wherein | A|=S, k=1,2 ... S, function C (J
1, J
2) return two image block J
1, J
2The number that equates gray-scale value, the i that makes the following formula establishment, (x y) is the starting point coordinate of detected original jpeg block boundary to the j value.
3, according to the altering detecting method of claims 1 described a kind of jpeg image, it is characterized in that described step 2) estimate that the specific practice of the little image block quantization table of JPEG is at first testing image to be sheared the capable y row of x, then the image that obtains is carried out piecemeal, dividing block size is B * B, for each image block J
i, estimate its quantization table, method of estimation is as follows: establishing jpeg image piece to be detected is J
i, the span of its quantization table is in set A, at first image block J
iCarry out the JPEG compression with the quantization table in the set A, obtain S width of cloth image, relatively this S width of cloth image and J
iThe number of equal gray-scale value, select therefrom to equate the maximum image block I of gray-scale value that the quantization table of I is then as image block J
iThe estimation of quantization table.
4, according to the altering detecting method of claims 1 described a kind of jpeg image, it is characterized in that described step 3) identification tampered image identifies by step 2 for utilizing different colours with the concrete grammar in positioning tampering zone) in detected different quantification table block, if it is inconsistent the color of large area region and background (the maximum piece of quantization table occurring) to occur, then think the jpeg image of being distorted.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200610036602 CN1916958A (en) | 2006-07-20 | 2006-07-20 | Method of detecting distorts of JPEG image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200610036602 CN1916958A (en) | 2006-07-20 | 2006-07-20 | Method of detecting distorts of JPEG image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN1916958A true CN1916958A (en) | 2007-02-21 |
Family
ID=37737950
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 200610036602 Pending CN1916958A (en) | 2006-07-20 | 2006-07-20 | Method of detecting distorts of JPEG image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1916958A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101710989B (en) * | 2009-11-06 | 2011-07-13 | 中山大学 | Method for identifying JPEG compressed digital picture |
CN102226920A (en) * | 2011-06-03 | 2011-10-26 | 贵州大学 | Cutting-resistant JPEG image compression history and synthetic tamper detection method |
CN102521821A (en) * | 2011-10-24 | 2012-06-27 | 南开大学 | Automatic identification and tampered area positioning method in digital image |
CN101674389B (en) * | 2009-09-30 | 2012-09-12 | 大连理工大学 | Method for testing compression history of BMP image based on loss amount of image information |
WO2012145904A1 (en) * | 2011-04-28 | 2012-11-01 | 中国科学院自动化研究所 | Method for tampered region localization of color images based on jpeg compression noise |
CN103839083A (en) * | 2014-02-25 | 2014-06-04 | 中山大学 | Method for detecting tampering operation of quantization tables of JPEG pictures |
CN104536843A (en) * | 2014-12-18 | 2015-04-22 | 无敌科技(西安)有限公司 | Embedded device and method for rapidly detecting whether JPEG picture files are right or not |
CN107464237A (en) * | 2017-08-04 | 2017-12-12 | 平安科技(深圳)有限公司 | Distorted image detection method, electronic installation and readable storage medium storing program for executing |
CN107948643A (en) * | 2017-11-27 | 2018-04-20 | 珠海亿智电子科技有限公司 | A kind of method for reducing jpeg image blocking artifact |
CN109544502A (en) * | 2018-09-25 | 2019-03-29 | 中山大学 | A kind of jpeg image down-sampling factor estimation method based on spectrum analysis and difference image extreme point range distribution |
-
2006
- 2006-07-20 CN CN 200610036602 patent/CN1916958A/en active Pending
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101674389B (en) * | 2009-09-30 | 2012-09-12 | 大连理工大学 | Method for testing compression history of BMP image based on loss amount of image information |
CN101710989B (en) * | 2009-11-06 | 2011-07-13 | 中山大学 | Method for identifying JPEG compressed digital picture |
WO2012145904A1 (en) * | 2011-04-28 | 2012-11-01 | 中国科学院自动化研究所 | Method for tampered region localization of color images based on jpeg compression noise |
CN102226920A (en) * | 2011-06-03 | 2011-10-26 | 贵州大学 | Cutting-resistant JPEG image compression history and synthetic tamper detection method |
CN102226920B (en) * | 2011-06-03 | 2013-04-17 | 贵州大学 | Cutting-resistant JPEG image compression history and synthetic tamper detection method |
CN102521821A (en) * | 2011-10-24 | 2012-06-27 | 南开大学 | Automatic identification and tampered area positioning method in digital image |
CN103839083A (en) * | 2014-02-25 | 2014-06-04 | 中山大学 | Method for detecting tampering operation of quantization tables of JPEG pictures |
CN103839083B (en) * | 2014-02-25 | 2017-10-20 | 中山大学 | The detection method that a kind of JPEG picture quantization table is distorted |
CN104536843A (en) * | 2014-12-18 | 2015-04-22 | 无敌科技(西安)有限公司 | Embedded device and method for rapidly detecting whether JPEG picture files are right or not |
CN107464237A (en) * | 2017-08-04 | 2017-12-12 | 平安科技(深圳)有限公司 | Distorted image detection method, electronic installation and readable storage medium storing program for executing |
CN107948643A (en) * | 2017-11-27 | 2018-04-20 | 珠海亿智电子科技有限公司 | A kind of method for reducing jpeg image blocking artifact |
CN107948643B (en) * | 2017-11-27 | 2020-10-23 | 珠海亿智电子科技有限公司 | Method for reducing block effect of JPEG image |
CN109544502A (en) * | 2018-09-25 | 2019-03-29 | 中山大学 | A kind of jpeg image down-sampling factor estimation method based on spectrum analysis and difference image extreme point range distribution |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1916958A (en) | Method of detecting distorts of JPEG image | |
US11263500B2 (en) | Image detection methods and apparatus | |
Yang et al. | An effective method for detecting double JPEG compression with the same quantization matrix | |
Li et al. | Rotation robust detection of copy-move forgery | |
Huang et al. | Improved DCT-based detection of copy-move forgery in images | |
Tralic et al. | CoMoFoD—New database for copy-move forgery detection | |
Ye et al. | Detecting digital image forgeries by measuring inconsistencies of blocking artifact | |
USRE40477E1 (en) | Reliable detection of LSB steganography in color and grayscale images | |
CN1184808C (en) | Embedding and detecting watermark in images | |
CN102184537B (en) | Image region tamper detection method based on wavelet transform and principal component analysis | |
Kumar et al. | A fast DCT based method for copy move forgery detection | |
Qu et al. | A convolutive mixing model for shifted double JPEG compression with application to passive image authentication | |
CN102957915B (en) | Double JPEG (Joint Photographic Experts Group) compressed image-targeted tamper detection and tamper locating method | |
CN110457996B (en) | Video moving object tampering evidence obtaining method based on VGG-11 convolutional neural network | |
Ketenci et al. | Copy-move forgery detection in images via 2D-Fourier transform | |
CN103561271A (en) | Video airspace tamper detection method for removing moving object shot by static camera lens | |
Khan et al. | Robust method for detection of copy-move forgery in digital images | |
Zhang et al. | A survey on passive-blind image forgery by doctor method detection | |
Muhammad et al. | A non-intrusive method for copy-move forgery detection | |
Fadl et al. | A proposed accelerated image copy-move forgery detection | |
CN107977964A (en) | Slit cropping evidence collecting method based on LBP and extension Markov feature | |
CN103561274B (en) | Video time domain tamper detection method for removing moving object shot by static camera lens | |
CN109660814B (en) | Method for detecting deletion tampering of video foreground | |
CN108269221B (en) | JPEG recompressed image tampering positioning method | |
CN1241389C (en) | Object-oriented self-synchronizing water mark detecting method for multimedia product |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |