CN102547371A - Secondary compression detection method based on H.264/AVC (Advanced Video Coding) video - Google Patents
Secondary compression detection method based on H.264/AVC (Advanced Video Coding) video Download PDFInfo
- Publication number
- CN102547371A CN102547371A CN2012100457427A CN201210045742A CN102547371A CN 102547371 A CN102547371 A CN 102547371A CN 2012100457427 A CN2012100457427 A CN 2012100457427A CN 201210045742 A CN201210045742 A CN 201210045742A CN 102547371 A CN102547371 A CN 102547371A
- Authority
- CN
- China
- Prior art keywords
- video
- compressed
- coefficient
- sequence
- compression
- 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.)
- Granted
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention belongs to the field of multimedia information safety and particularly relates to a secondary compression detection method based on an H.264/AVC (Advanced Video Coding) video. The method comprises the following steps of: carrying out primary compression on an uncompressed video sample plate sequence according to second quantization parameters to obtain a primary compressed video library; after carrying out the primary compression on the uncompressed video sample plate sequence according to first quantization parameters, decompressing and carrying out secondary compression according to second quantization parameters to obtain a secondary compressed video library; extracting feature vectors of a primary compressed video sequence and a secondary compressed video sequence; utilizing a classifier to train the feature vectors of the primary compressed video sequence and the secondary compressed video sequence to obtain a classification training model; extracting the feature vectors of videos to be detected and utilizing the classification training model to perform prediction and classification to divide the videos to be detected into two classes, namely primary compressed videos and secondary compressed videos. According to the secondary compression detection method provided by the invention, the videos subjected to the secondary compression can be effectively detected.
Description
Technical field
The invention belongs to the multi-media information security field, be specifically related to a kind of based on video second-compressed detection method H.264/AVC.
Background technology
Popularizing of the progress of Along with computer technology and Video processing software (like meeting sound meeting shadow, Adobe Premier etc.), domestic consumer is also more and more easier to the operation of video.If the assailant carries out malice to video and distorts, for example will contain the frame deletion of key message and the frame that insertion is distorted, all can influence video content, cause the user that deviation is appearred in factual judgement simultaneously.Therefore under situations such as court's evidence obtaining, Digital Media copyright protection, the integrality of checking video content becomes more and more urgent.On the other hand, for conserve storage and bandwidth, most of videos are all with compressed code flow stored in form or distribution.In the source of video record, recording arrangement (first-class like hand-held DV, monitoring camera) all is integrated into compression function on the chip, compresses while record, then with the code stream stored in form.For the assailant, distort in pixel domain usually and distort more easily than code stream, therefore at first need video be decompressed to pixel domain, distort attack after, recompress into code stream to preserve.This just means that video has passed through twice compression, and accomplished by hardware device the first time in video record, is accomplished by the assailant for the second time.If can detect video has experienced twice compression, a so such video is suspicious, and its content possibly be incomplete.
H.264/AVC be based on infra-frame prediction for the coding of I frame; To block to be encoded; Choose adjacent encoding block piece as a reference, obtain the optimum prediction macro block, deduct predicted macroblock with original macro to be encoded and obtain the residual error between the two through certain predictive mode; And then residual signals carried out dct transform, quantification, entropy coding, finally write code stream.H.264/AVC compression standard owing to its excellent compression performance and good network affinity, becomes the main flow compress mode of video compression after becoming international standard in 2003.Therefore to being detected as for an emphasis based on the second-compressed of video H.264/AVC.
Summary of the invention
The technical problem that the present invention solves is the deficiency that overcomes prior art, provide a kind of can effectively detect through the video of second-compressed based on video second-compressed detection method H.264/AVC.
For solving the problems of the technologies described above, technical scheme of the present invention is following:
A kind of based on video second-compressed detection method H.264/AVC, comprise the steps:
(1) set up second-compressed visual classification training pattern, its specifically:
(11) unpressed video model sequence is once compressed respectively with second-compressed obtained compressed video storehouse and second-compressed video library;
(12) respectively compression of video sequence in the compressed video storehouse and the second-compressed video sequence in the second-compressed video library are carried out the entropy decoding, summation about non-zero DCT coefficients after the I frame that obtains each compression of video sequence and each second-compressed video sequence quantizes and the DCT coefficient behind the inverse quantization;
(13) extract the set of eigenvectors of compression of video sequence and second-compressed video sequence respectively according to the summation about non-zero DCT coefficients after the I frame quantification of compression of video sequence that obtains and second-compressed video sequence and the DCT coefficient behind the inverse quantization;
(14) utilize grader that the set of eigenvectors of compression of video sequence and second-compressed video sequence is trained and obtain the classification based training model;
(2) utilize the classification based training model that video to be detected is carried out the second-compressed Video Detection, its specifically:
(21) extract the characteristic vector of video to be measured, utilize the classification based training model to predict classification, video to be measured is divided into two types: compressed video and second-compressed video.
In the such scheme, the concrete steps of said step (11) are:
(111) unpressed video model sequence is carried out H.264/AVC compressed video storehouse of compression acquisition one time with second quantization parameter;
(112) unpressed video model sequence is carried out H.264/AVC compressing the first time with first quantization parameter after, decompress(ion) carries out H.264/AVC compressing the second time with second quantization parameter again and obtains the second-compressed video library; Said first quantization parameter is less than second quantization parameter.
In the such scheme, the concrete steps of said step (13) are:
(131) coefficient threshold range and frequency threshold scope are set;
(132) probability distribution of DC coefficient absolute value in the coefficient threshold range in the summation about non-zero DCT coefficients after the I frame that calculates each compression of video sequence and each second-compressed video sequence quantizes;
(133) ac coefficient in the summation about non-zero DCT coefficients medium frequency threshold range after the I frame that extracts each compression of video sequence and each second-compressed video sequence quantizes, the probability distribution of ac coefficient absolute value in the coefficient threshold range in the calculated rate threshold range;
(134) probability distribution of ac coefficient absolute value in the coefficient threshold range in the summation about non-zero DCT coefficients after the I frame that calculates each compression of video sequence and each second-compressed video sequence quantizes on all frequencies;
(135) the DCT coefficient histogram behind the I frame inverse quantization of each compression of video sequence and each second-compressed video sequence is carried out FFT, obtain energy spectral density;
(136) extract maximum point and minimum point in the energy spectral density of each compression of video sequence and each second-compressed video sequence; Choose an equal number maximum maximum point and minimum minimum point; With the frequency distribution value constitutive characteristic vector set that is obtained in value of choosing and step (132)-(134), obtain the set of eigenvectors of compression of video sequence and second-compressed video sequence.
In the such scheme, extract the characteristic vector of video to be measured in the said step (21), utilize the classification based training model to predict classification, video to be measured is divided into two types: the concrete steps of compressed video and second-compressed video are:
(211) video to be detected is carried out entropy decoding, obtain summation about non-zero DCT coefficients and the DCT coefficient behind the inverse quantization after the I frame quantizes;
(211) coefficient threshold range and frequency threshold scope are set;
(212) probability distribution of DC coefficient absolute value in the coefficient threshold range in the summation about non-zero DCT coefficients after the I frame of calculating video to be detected quantizes;
(213) ac coefficient in the summation about non-zero DCT coefficients medium frequency threshold range after the I frame of extraction video to be detected quantizes, the probability distribution of ac coefficient absolute value in the coefficient threshold range in the calculated rate threshold range;
(214) probability distribution of ac coefficient absolute value in the coefficient threshold range in the summation about non-zero DCT coefficients after the I frame of calculating video to be detected quantizes on all frequencies;
(215) the DCT coefficient histogram behind the I frame inverse quantization of video to be detected is carried out FFT, obtain energy spectral density;
(216) maximum point and the minimum point in the energy spectral density of extraction video to be detected chosen the characteristic vector that an equal number maximum maximum point and minimum minimum point constitute video to be detected;
(217) predict classification in the characteristic vector input category training pattern according to the video to be detected that obtains, video to be measured is divided into two types: compressed video and second-compressed video.
In the such scheme, said coefficient threshold range is set to [1,10], and said frequency threshold scope is set to (0,1) and (1,0).
In the such scheme, said video model sequence is the yuv video sequence.
In the such scheme, said grader is a SVMs.
Compared with prior art, the beneficial effect of technical scheme of the present invention is:
The present invention is the statistic descriminant technique that proposes according to video compression standard intraframe coding method H.264/AVC; It possibly pass through second-compressed to the video after distorting; Analyze the probability distribution situation that the I frame quantizes the ac coefficient of back non-zero ac coefficient, DC coefficient and CF, simultaneously the histogrammic energy spectral density of DCT coefficient is analyzed, extract correlated characteristic; Utilize the support vector machine technology to classify, the integrality and the authenticity of video are differentiated.The present invention can effectively detect the video of second-compressed, is particularly useful for compressing the rear video mass ratio for the second time and compresses the low-quality situation of rear video for the first time.
Description of drawings
General video is distorted the flow chart of attack in Fig. 1 prior art;
Fig. 2 is a flow chart of the present invention;
Fig. 3 is the compressed video utilizing in the specific embodiment among the present invention that different quantization parameter obtains and the energy spectral density design sketch of second-compressed video.
Embodiment
Below in conjunction with accompanying drawing and embodiment technical scheme of the present invention is done further explanation.
It is a kind of based on the flow chart of video second-compressed detection method H.264/AVC to be illustrated in figure 2 as the present invention, and it specifically comprises the steps:
(S1) set up second-compressed visual classification training pattern, its specifically:
(S11) unpressed video model sequence is once compressed respectively with second-compressed obtained compressed video storehouse and second-compressed video library; It is specifically:
(S111) unpressed video model sequence is carried out H.264/AVC compressed video storehouse of compression acquisition one time with second quantization parameter; Video model sequence adopts the yuv video sequence, and each yuv video sequence is done H.264/AVC compression with the second quantization parameter QP2, obtains the compressed video storehouse one time.
(S112) unpressed video model sequence is carried out H.264/AVC compressing the first time with first quantization parameter after, decompress(ion) carries out H.264/AVC compressing the second time with second quantization parameter again and obtains the second-compressed video library; Said first quantization parameter is less than second quantization parameter.Unpressed video model sequence adopts the uncompressed video model sequence in the step (S111); Step (S112) is at first done a H.264/AVC compression with the first quantization parameter QP1 to each yuv video sequence; Carry out decompress(ion) then; Make secondary with the second quantization parameter QP2 again behind the decompress(ion) and H.264/AVC compress, guarantee that the first quantization parameter QP1 less than the second quantization parameter QP2, obtains the second-compressed video library.
(S12) respectively compression of video sequence in the compressed video storehouse and the second-compressed video sequence in the second-compressed video library are carried out the entropy decoding, summation about non-zero DCT coefficients after the I frame that obtains each compression of video sequence and each second-compressed video sequence quantizes and the DCT coefficient behind the inverse quantization; It specifically is each compression of video sequence to be carried out entropy decoding obtain summation about non-zero DCT coefficients and the DCT coefficient behind the inverse quantization after the I frame of each compression of video sequence quantizes, and it is inferior that each second-compressed video sequence is carried out summation about non-zero DCT coefficients and the DCT coefficient behind the inverse quantization that the entropy decoding obtains after the I frame of each second-compressed video sequence quantizes.
(S13) extract the set of eigenvectors of compression of video sequence and second-compressed video sequence respectively according to the summation about non-zero DCT coefficients after the I frame quantification of compression of video sequence that obtains and second-compressed video sequence and the DCT coefficient behind the inverse quantization; Its concrete steps are:
(S131) coefficient threshold range and frequency threshold scope are set; The coefficient threshold range is set to [1,10], and the frequency threshold scope is set to (0,1) and (1,0).
(S132) probability distribution of DC coefficient absolute value in coefficient threshold range [1,10] in the summation about non-zero DCT coefficients after the I frame that calculates each compression of video sequence and each second-compressed video sequence respectively quantizes;
(S133) the summation about non-zero DCT coefficients medium frequency threshold range (0 after the I frame that extracts each compression of video sequence and each second-compressed video sequence respectively quantizes; 1) reaches (1; 0) ac coefficient in; Distinguish the probability distribution of ac coefficient absolute value in coefficient threshold range [1,10] in the calculated rate threshold range then;
(S134) probability distribution of ac coefficient absolute value in coefficient threshold range [1,10] in the summation about non-zero DCT coefficients after the I frame that calculates each compression of video sequence and each second-compressed video sequence respectively quantizes on all frequencies;
(S135) the DCT coefficient histogram behind the I frame inverse quantization of each compression of video sequence and each second-compressed video sequence is carried out FFT respectively, obtain the energy spectral density of each compression of video sequence and each second-compressed video sequence;
(S136) extract maximum point and minimum point in the energy spectral density of each compression of video sequence and each second-compressed video sequence respectively; Choose an equal number maximum maximum point and minimum minimum point constitutive characteristic vector; As choose the minimum point constitutive characteristic vector of maximum point and 3 minimums of 3 maximums, obtain the set of eigenvectors of compression of video sequence and second-compressed video sequence; The probability distribution of calculating in the step (S132)-(S134); Therefore each step obtains 10 frequency distribution values in scope [1,10], totally 30 dimensional features; Add the maximum point of 3 maximums and the minimum point of 3 minimums; Constitute the characteristic vector of 36 dimensions of a compression of video sequence, and the characteristic vector of 36 dimensions of second-compressed video sequence, latter two characteristic vector constitutive characteristic vector set.
(S14) utilize grader that the set of eigenvectors of compression of video sequence and second-compressed video sequence is trained and obtain the classification based training model; Grader can adopt SVMs.
(S2) utilize the classification based training model that video to be detected is carried out the second-compressed Video Detection, its specifically:
(S21) extract the characteristic vector of video to be measured, utilize the classification based training model to predict classification, video to be measured is divided into two types: compressed video and second-compressed video; Its concrete steps are:
(S211) video to be detected is carried out entropy decoding, obtain summation about non-zero DCT coefficients and the DCT coefficient behind the inverse quantization after the I frame quantizes;
(S212) coefficient threshold range and frequency threshold scope are set; The coefficient threshold range is set to [1,10], and the frequency threshold scope is set to (0,1) and (1,0).
(S213) probability distribution of DC coefficient absolute value in coefficient threshold range [1,10] in the summation about non-zero DCT coefficients after the I frame of calculating video to be detected quantizes;
(S214) the summation about non-zero DCT coefficients medium frequency threshold range (0,1) after the I frame of extraction video to be detected quantizes and the ac coefficient in (1,0), the ac coefficient absolute value in the calculated rate threshold range is in the interior probability distribution of coefficient threshold range [1,10];
(S215) probability distribution of ac coefficient absolute value in coefficient threshold range [1,10] in the summation about non-zero DCT coefficients after the I frame of calculating video to be detected quantizes on all frequencies;
(S216) the DCT coefficient histogram behind the I frame inverse quantization of video to be detected is carried out FFT respectively, obtain the energy spectral density of video to be detected;
(S217) maximum point and the minimum point in the energy spectral density of extraction video to be detected; Choose the maximum maximum point of equal number and constitute the characteristic vector of video to be detected, as the minimum point of choosing maximum point and 3 minimums of 3 maximums constitutes the characteristic vector of video to be detected with minimum minimum point;
(S218) predict classification in the characteristic vector input category training pattern according to the video to be detected that obtains, video to be measured is divided into two types: compressed video and second-compressed video.
Below in conjunction with concrete embodiment method of the present invention is specified, be to be noted that described embodiment is intended to be convenient to understanding of the present invention is not played the qualification effect to the present invention.
At first constructing the compressed video storehouse one time, choose the yuv video sequence of 30 1665 frames and once compress with the X264 encoder, in order to increase the quantity of video library, is that unit is cut apart with 30 yuv video sequences with 10 frames, and promptly each subsequence contains 10 frames.Each subsequence is compressed, choose the second quantization parameter QP2 for 22,24,26,28,30}.
Next constructs the second-compressed video library; Each yuv video subsequence is at first once compressed with the first quantization parameter QP1, carry out second-compressed with the second quantization parameter QP2 again behind the decompress(ion), wherein QP2 is greater than QP1; The employed second quantization parameter QP2 in compressed video storehouse is the same for QP2 value and structure; Be 22,24,26,28,30, the QP1 span is QP1=QP2-x, x=1,2,3,4,5.
The extraction of characteristic vector; Each yuv video subsequence in compressed video storehouse and the second-compressed video library is carried out the entropy decoding; DCT coefficient behind non-zero ac coefficient, DC coefficient and the inverse quantization after the I frame that extracts each yuv video subsequence quantizes, the DCT coefficient behind non-zero ac coefficient, DC coefficient and the inverse quantization after quantizing according to the I frame of each yuv video subsequence of extracting are carried out the set of eigenvectors that said step (S141)-(S146) extracts compression of video sequence and second-compressed video sequence respectively.
The structural classification device utilizes SVMs svm classifier device that two set of eigenvectors that obtain are trained, and obtains one and can judge < whether video has passed through the classification based training model of second-compressed under the situation of QP2 at QP1.
Video to be detected is carried out second-compressed to be detected; Video to be detected is carried out the entropy decoding; DCT coefficient behind non-zero ac coefficient, DC coefficient and the inverse quantization after the I frame that extracts video to be detected quantizes; Carry out the characteristic vector that said step (S131)-(S136) extracts video to be detected, be input in the classification based training model and train differentiation, distinguish compressed video and second-compressed video with this.Experimental result is as shown in table 1 below:
In last table, be to utilize the embodiment of the invention to set up the classification based training model 100 videos to be detected are carried out the testing result of secondary detection, can know that from last table utilizing the present invention that video is carried out the accuracy rate that second-compressed detects can reach more than 80%.As shown in Figure 3; The 1st curve is represented the energy spectral density curve of a compression of video sequence, and second quantization parameter that uses is QP2=26, and other curves are represented the energy spectral density curve of second-compressed video sequence; First quantization parameter is respectively QP1=21,23,25; Second quantization parameter is QP2=26, and as can beappreciated from fig. 3, the numerical value of the extreme point of the energy spectral density of second-compressed video sequence all is higher than the analog value that compression of video sequence obtains.Therefore can utilize extreme point in the energy spectral density to distinguish the video of once compression and second-compressed.The present invention can effectively detect the video of second-compressed, is particularly useful for compressing the rear video mass ratio for the second time and compresses the low-quality situation of rear video for the first time.
Claims (7)
1. one kind based on video second-compressed detection method H.264/AVC, it is characterized in that, comprises the steps:
(1) set up second-compressed visual classification training pattern, its specifically:
(11) unpressed video model sequence is once compressed respectively with second-compressed obtained compressed video storehouse and second-compressed video library;
(12) respectively compression of video sequence in the compressed video storehouse and the second-compressed video sequence in the second-compressed video library are carried out the entropy decoding, summation about non-zero DCT coefficients after the I frame that obtains each compression of video sequence and each second-compressed video sequence quantizes and the DCT coefficient behind the inverse quantization;
(13) extract the set of eigenvectors of compression of video sequence and second-compressed video sequence respectively according to the summation about non-zero DCT coefficients after the I frame quantification of compression of video sequence that obtains and second-compressed video sequence and the DCT coefficient behind the inverse quantization;
(14) utilize grader that the set of eigenvectors of compression of video sequence and second-compressed video sequence is trained and obtain the classification based training model;
(2) utilize the classification based training model that video to be detected is carried out the second-compressed Video Detection, its specifically:
(21) extract the characteristic vector of video to be measured, utilize the classification based training model to predict classification, video to be measured is divided into two types: compressed video and second-compressed video.
2. according to claim 1 based on video second-compressed detection method H.264/AVC, it is characterized in that the concrete steps of said step (11) are:
(111) unpressed video model sequence is carried out H.264/AVC compressed video storehouse of compression acquisition one time with second quantization parameter;
(112) unpressed video model sequence is carried out H.264/AVC compressing the first time with first quantization parameter after, decompress(ion) carries out H.264/AVC compressing the second time with second quantization parameter again and obtains the second-compressed video library; Said first quantization parameter is less than second quantization parameter.
3. according to claim 2 based on video second-compressed detection method H.264/AVC, it is characterized in that said step (13) concrete steps are:
(131) coefficient threshold range and frequency threshold scope are set;
(132) probability distribution of DC coefficient absolute value in the coefficient threshold range in the summation about non-zero DCT coefficients after the I frame that calculates each compression of video sequence and each second-compressed video sequence quantizes;
(133) ac coefficient in the summation about non-zero DCT coefficients medium frequency threshold range after the I frame that extracts each compression of video sequence and each second-compressed video sequence quantizes, the probability distribution of ac coefficient absolute value in the coefficient threshold range in the calculated rate threshold range;
(134) probability distribution of ac coefficient absolute value in the coefficient threshold range in the summation about non-zero DCT coefficients after the I frame that calculates each compression of video sequence and each second-compressed video sequence quantizes on all frequencies;
(135) the DCT coefficient histogram behind the I frame inverse quantization of each compression of video sequence and each second-compressed video sequence is carried out FFT, obtain energy spectral density;
(136) extract maximum point and minimum point in the energy spectral density of each compression of video sequence and each second-compressed video sequence; Choose an equal number maximum maximum point and minimum minimum point; With the frequency distribution value constitutive characteristic that obtained in value of choosing and step (132)-(134) vector, obtain the set of eigenvectors of compression of video sequence and second-compressed video sequence.
4. according to claim 3 based on video second-compressed detection method H.264/AVC; It is characterized in that; Extract the characteristic vector of video to be measured in the said step (21); Utilize the classification based training model to predict classification, video to be measured is divided into two types: the concrete steps of compressed video and second-compressed video are:
(211) video to be detected is carried out entropy decoding, obtain summation about non-zero DCT coefficients and the DCT coefficient behind the inverse quantization after the I frame quantizes;
(211) coefficient threshold range and frequency threshold scope are set;
(212) probability distribution of DC coefficient absolute value in the coefficient threshold range in the summation about non-zero DCT coefficients after the I frame of calculating video to be detected quantizes;
(213) ac coefficient in the summation about non-zero DCT coefficients medium frequency threshold range after the I frame of extraction video to be detected quantizes, the probability distribution of ac coefficient absolute value in the coefficient threshold range in the calculated rate threshold range;
(214) probability distribution of ac coefficient absolute value in the coefficient threshold range in the summation about non-zero DCT coefficients after the I frame of calculating video to be detected quantizes on all frequencies;
(215) the DCT coefficient histogram behind the I frame inverse quantization of video to be detected is carried out FFT, obtain energy spectral density;
(216) maximum point and the minimum point in the energy spectral density of extraction video to be detected chosen the characteristic vector that an equal number maximum maximum point and minimum minimum point constitute video to be detected;
(217) predict classification in the characteristic vector input category training pattern according to the video to be detected that obtains, video to be measured is divided into two types: compressed video and second-compressed video.
5. each describedly is characterized in that based on video second-compressed detection method H.264/AVC said coefficient threshold range is set to [1,10] according to claim 1 to 4, and said frequency threshold scope is set to (0,1) and (1,0).
6. each is described based on video second-compressed detection method H.264/AVC according to claim 1 to 4, it is characterized in that said video model sequence is the yuv video sequence.
7. each is described based on video second-compressed detection method H.264/AVC according to claim 1 to 4, it is characterized in that said grader is a SVMs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210045742.7A CN102547371B (en) | 2012-02-27 | 2012-02-27 | Secondary compression detection method based on H.264/AVC (Advanced Video Coding) video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210045742.7A CN102547371B (en) | 2012-02-27 | 2012-02-27 | Secondary compression detection method based on H.264/AVC (Advanced Video Coding) video |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102547371A true CN102547371A (en) | 2012-07-04 |
| CN102547371B CN102547371B (en) | 2014-02-19 |
Family
ID=46353128
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210045742.7A Expired - Fee Related CN102547371B (en) | 2012-02-27 | 2012-02-27 | Secondary compression detection method based on H.264/AVC (Advanced Video Coding) video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102547371B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103607592A (en) * | 2013-11-28 | 2014-02-26 | 广东威创视讯科技股份有限公司 | Video tampering detection method and system |
| CN104837028A (en) * | 2015-04-29 | 2015-08-12 | 上海交通大学 | Video same-bit-rate dual-compression detection method |
| CN105448299A (en) * | 2015-11-17 | 2016-03-30 | 中山大学 | Method of discriminating digital audio AAC format codec |
| CN106331730A (en) * | 2016-08-22 | 2017-01-11 | 上海交通大学 | Double-compression detection method by using quantification factor same as H.264 video |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008005950A2 (en) * | 2006-06-30 | 2008-01-10 | New Jersey Institute Of Technology | An apparatus and method for a generalized benford's law for analysis of dct and jpeg coefficients |
| US20090263011A1 (en) * | 2008-04-18 | 2009-10-22 | Yun-Qing Shi | Detection Technique for Digitally Altered Images |
| CN101790097A (en) * | 2010-03-05 | 2010-07-28 | 天津大学 | Method for detecting multiple times of compression and coding of digital video |
| CN101989355A (en) * | 2009-07-31 | 2011-03-23 | 北京邮电大学 | Detection method for secondary compression of JPEG image |
| CN102226920A (en) * | 2011-06-03 | 2011-10-26 | 贵州大学 | Cutting-resistant JPEG image compression history and synthetic tamper detection method |
-
2012
- 2012-02-27 CN CN201210045742.7A patent/CN102547371B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008005950A2 (en) * | 2006-06-30 | 2008-01-10 | New Jersey Institute Of Technology | An apparatus and method for a generalized benford's law for analysis of dct and jpeg coefficients |
| US20090263011A1 (en) * | 2008-04-18 | 2009-10-22 | Yun-Qing Shi | Detection Technique for Digitally Altered Images |
| CN101989355A (en) * | 2009-07-31 | 2011-03-23 | 北京邮电大学 | Detection method for secondary compression of JPEG image |
| CN101790097A (en) * | 2010-03-05 | 2010-07-28 | 天津大学 | Method for detecting multiple times of compression and coding of digital video |
| CN102226920A (en) * | 2011-06-03 | 2011-10-26 | 贵州大学 | Cutting-resistant JPEG image compression history and synthetic tamper detection method |
Non-Patent Citations (3)
| Title |
|---|
| 戴蒙: "JPEG二次压缩的分析与检测", 《中国图象图形学报》 * |
| 戴蒙: "基于JPEG二次压缩的隐藏信息检测", 《华东理工大学学报(自然科学版)》 * |
| 韩晓东: "一种新的JPEG图像二次压缩检测算法", 《计算机工程》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103607592A (en) * | 2013-11-28 | 2014-02-26 | 广东威创视讯科技股份有限公司 | Video tampering detection method and system |
| CN103607592B (en) * | 2013-11-28 | 2017-02-15 | 广东威创视讯科技股份有限公司 | Video tampering detection method and system |
| CN104837028A (en) * | 2015-04-29 | 2015-08-12 | 上海交通大学 | Video same-bit-rate dual-compression detection method |
| CN104837028B (en) * | 2015-04-29 | 2017-12-15 | 上海交通大学 | Video is the same as bit rate dual compression detection method |
| CN105448299A (en) * | 2015-11-17 | 2016-03-30 | 中山大学 | Method of discriminating digital audio AAC format codec |
| CN105448299B (en) * | 2015-11-17 | 2019-04-05 | 中山大学 | A method of identifying digital audio AAC format codec |
| CN106331730A (en) * | 2016-08-22 | 2017-01-11 | 上海交通大学 | Double-compression detection method by using quantification factor same as H.264 video |
| CN106331730B (en) * | 2016-08-22 | 2020-03-06 | 上海交通大学 | H.264 video same quantization factor double compression detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102547371B (en) | 2014-02-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP4566498B2 (en) | How to describe motion activity in video | |
| Zhai et al. | Universal detection of video steganography in multiple domains based on the consistency of motion vectors | |
| Milani et al. | Multiple compression detection for video sequences | |
| Su et al. | Detection of double-compression in MPEG-2 videos | |
| Wang et al. | Real-time compressed-domain video watermarking resistance to geometric distortions | |
| CN101605272B (en) | Method for evaluating objective quality of partial reference type image | |
| CN102547371B (en) | Secondary compression detection method based on H.264/AVC (Advanced Video Coding) video | |
| He et al. | Double compression detection based on local motion vector field analysis in static-background videos | |
| Jiang et al. | Detection of HEVC double compression with the same coding parameters based on analysis of intra coding quality degradation process | |
| CN103067713B (en) | Method and system of bitmap joint photographic experts group (JPEG) compression detection | |
| KR101149522B1 (en) | Apparatus and method for detecting scene change | |
| CN109982071B (en) | HEVC (high efficiency video coding) dual-compression video detection method based on space-time complexity measurement and local prediction residual distribution | |
| Bakas et al. | Mpeg double compression based intra-frame video forgery detection using cnn | |
| Li et al. | A semi-fragile video watermarking algorithm based on H. 264/AVC | |
| CN104853186A (en) | Improved video steganalysis method based on motion vector reply | |
| Bagdanov et al. | Adaptive video compression for video surveillance applications | |
| CN103020138A (en) | Method and device for video retrieval | |
| CN102595136A (en) | H.264/AVC (automatic volume control) video secondary compression detection method based on quantized coefficient statistical properties | |
| CN105915916B (en) | Video steganalysis method based on the estimation of motion vector distortion performance | |
| Xu et al. | Detection of video transcoding for digital forensics | |
| Zhang et al. | Detecting multiple H. 264/AVC compressions with the same quantisation parameters | |
| Neufeld et al. | A study of embedding operations and locations for steganography in H. 264 video | |
| Ikusan et al. | Rate-distortion optimized hierarchical deep feature compression | |
| Wu et al. | HEVC double compression detection with non-aligned GOP structures based on a fusion feature with optical flow and prediction units | |
| Chu et al. | Forensic identification of compressively sensed images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140219 Termination date: 20180227 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |