CN102547371B - 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
- CN102547371B CN102547371B CN201210045742.7A CN201210045742A CN102547371B CN 102547371 B CN102547371 B CN 102547371B CN 201210045742 A CN201210045742 A CN 201210045742A CN 102547371 B CN102547371 B CN 102547371B
- Authority
- CN
- China
- Prior art keywords
- video
- compressed
- compression
- coefficient
- sequence
- 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.)
- Expired - Fee Related
Links
- 230000006835 compression Effects 0.000 title claims abstract description 89
- 238000007906 compression Methods 0.000 title claims abstract description 89
- 238000001514 detection method Methods 0.000 title claims abstract description 19
- 238000013139 quantization Methods 0.000 claims abstract description 51
- 239000013598 vector Substances 0.000 claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 22
- 239000000284 extract Substances 0.000 claims description 25
- 230000003595 spectral effect Effects 0.000 claims description 18
- 238000012706 support-vector machine Methods 0.000 claims description 5
- 230000000007 visual effect Effects 0.000 claims description 3
- 238000000034 method Methods 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
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 multi-media information security field, be specifically related to a kind of based on video second-compressed detection method H.264/AVC.
Background technology
Along with the progress of computer technology and popularizing of Video processing software (as meeting sound meeting shadow, Adobe Premier etc.), domestic consumer is also more and more easier to the operation of video.If assailant carries out malice to video, distort, the frame of for example frame deletion that contains key message and insertion being distorted, all can affect video content, causes user to occur deviation to true judgement simultaneously.Therefore under the 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, in order to save memory space and bandwidth, most of videos are all with the storage of code stream form or distribution after compression.In the source of video record, recording arrangement (as first-class in hand-held DV, monitoring camera) is all integrated into compression function on chip, while record, compresses, and then with code stream form, stores.For assailant, conventionally in pixel domain, distort and than code stream, distort more easily, therefore first need video to decompress to pixel domain, after Tampering attack, recompress into code stream to preserve.This just means that video has passed through twice compression, is completed for the first time in video record by hardware device, is completed for the second time by assailant.If video can be detected, experienced twice compression, so such a video is suspicious, and its content may be incomplete.
H.264/AVC for the coding of I frame based on infra-frame prediction, to block to be encoded, choose adjacent coded block as with reference to piece, by certain predictive mode, obtain optimum prediction macro block, by original macro to be encoded, deduct predicted macroblock and obtain the residual error between the two, and then residual signals is carried out to dct transform, quantification, entropy coding, finally write code stream.H.264/AVC compression standard became after international standard since 2003, due to its excellent compression performance and good network affinity, became the main flow compress mode of video compression.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 to overcome the deficiencies in the 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 as follows:
H.264/AVC a video second-compressed detection method, comprises the steps:
(1) set up second-compressed visual classification training pattern, its specifically:
(11) unpressed video model sequence is carried out respectively to first compression and second-compressed and obtain first compression video library and second-compressed video library;
(12) respectively the first compression video sequence in first compression video library and the second-compressed video sequence in second-compressed video library are carried out to entropy decoding, the summation about non-zero DCT coefficients after the I frame that obtains each first compression video sequence and each second-compressed video sequence quantizes and the DCT coefficient after inverse quantization;
(13) set of eigenvectors that the summation about non-zero DCT coefficients after the first compression video sequence that basis is obtained and the I frame of second-compressed video sequence quantize and the DCT coefficient after inverse quantization extract respectively first compression video sequence and second-compressed video sequence;
(14) utilize grader to train and obtain classification based training model the set of eigenvectors of first compression video sequence and second-compressed video sequence;
(2) utilize classification based training model to carry out the detection of second-compressed video to video to be detected, its specifically:
(21) extract the characteristic vector of video to be measured, utilize classification based training model to predict classification, video to be measured is divided into two classes: first compression video and second-compressed video.
In such scheme, the concrete steps of described step (11) are:
(111) unpressed video model sequence is carried out to H.264/AVC compression acquisition first compression video library one time with the second quantization parameter;
(112) after H.264/AVC unpressed video model sequence is compressed for the first time with the first quantization parameter, decompress(ion), more H.264/AVC compress for the second time and obtain second-compressed video library with the second quantization parameter; Described the first quantization parameter is less than the second quantization parameter.
In such scheme, the concrete steps of described step (13) are:
(131) coefficient threshold range and frequency threshold scope are set;
(132) probability distribution of DC coefficient absolute value in coefficient threshold range in the summation about non-zero DCT coefficients after the I frame that calculates each first compression 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 first compression video sequence and each second-compressed video sequence quantizes, the probability distribution of the ac coefficient absolute value in calculated rate threshold range in coefficient threshold range;
(134) probability distribution of the ac coefficient absolute value in the summation about non-zero DCT coefficients after the I frame that calculates each first compression video sequence and each second-compressed video sequence quantizes in all frequencies in coefficient threshold range;
(135) the DCT coefficient histogram after the I frame inverse quantization of each first compression video sequence and each second-compressed video sequence is carried out to fast fourier transform, obtain energy spectral density;
(136) extract maximum point and the minimum point in the energy spectral density of each first compression video sequence and each second-compressed video sequence, choose an equal number maximum maximum point and minimum minimum point, by the frequency distribution value constitutive characteristic vector set obtaining in the value of choosing and step (132)-(134), obtain the set of eigenvectors of first compression video sequence and second-compressed video sequence.
In such scheme, in described step (21), extract the characteristic vector of video to be measured, utilize classification based training model to predict classification, video to be measured is divided into two classes: the concrete steps of first compression video and second-compressed video are:
(211) video to be detected is carried out to entropy decoding, the summation about non-zero DCT coefficients after obtaining I frame and quantizing and the DCT coefficient after inverse quantization;
(211) coefficient threshold range and frequency threshold scope are set;
(212) probability distribution of DC coefficient absolute value in coefficient threshold range in the summation about non-zero DCT coefficients after the I frame that calculates video to be detected quantizes;
(213) ac coefficient in the summation about non-zero DCT coefficients medium frequency threshold range after the I frame that extracts video to be detected quantizes, the probability distribution of the ac coefficient absolute value in calculated rate threshold range in coefficient threshold range;
(214) probability distribution of the ac coefficient absolute value in the summation about non-zero DCT coefficients after the I frame that calculates video to be detected quantizes in all frequencies in coefficient threshold range;
(215) the DCT coefficient histogram after the I frame inverse quantization of video to be detected is carried out to fast fourier transform, obtain energy spectral density;
(216) extract maximum point and the minimum point in the energy spectral density of video to be detected, choose the characteristic vector that an equal number maximum maximum point and minimum minimum point form video to be detected;
(217) according to predicting classification in the characteristic vector input classification based training model of the video to be detected obtaining, video to be measured is divided into two classes: first compression video and second-compressed video.
In such scheme, described coefficient threshold range is set to [1,10], and described frequency threshold scope is set to (0,1) and (1,0).
In such scheme, described video model sequence is yuv video sequence.
In such scheme, described grader is SVMs.
Compared with prior art, the beneficial effect of technical solution of the present invention is:
The present invention is the statistic descriminant technique according to H.264/AVC video compression standard intraframe coding method proposes, it may pass through second-compressed for the video after distorting, analyze the probability distribution situation that I frame quantizes the ac coefficient of rear non-zero ac coefficient, DC coefficient and characteristic frequency, the histogrammic energy spectral density of DCT coefficient is analyzed simultaneously, extract correlated characteristic, utilize support vector machine technology to classify, the integrality of video and authenticity are differentiated.The present invention can effectively detect the video of second-compressed, is particularly useful for compressing for the second time rear video mass ratio and compresses for the first time the low-quality situation of rear video.
Accompanying drawing explanation
The flow chart of general video Tampering attack in Fig. 1 prior art;
Fig. 2 is flow chart of the present invention;
Fig. 3 utilizes first compression video that different quantization parameters obtain and the energy spectral density design sketch of second-compressed video in specific embodiment in the present invention.
Embodiment
Below in conjunction with drawings and Examples, technical scheme of the present invention is described further.
Be illustrated in figure 2 the present invention a kind of based on the flow chart of video second-compressed detection method H.264/AVC, it specifically comprises the steps:
(S1) set up second-compressed visual classification training pattern, its specifically:
(S11) unpressed video model sequence is carried out respectively to first compression and second-compressed and obtain first compression video library and second-compressed video library; It is specifically:
(S111) unpressed video model sequence is carried out to H.264/AVC compression acquisition first compression video library one time with the second quantization parameter; Video model sequence adopts yuv video sequence, and each yuv video sequence is done to H.264/AVC compression with the second quantization parameter QP2, obtains first compression video library.
(S112) after H.264/AVC unpressed video model sequence is compressed for the first time with the first quantization parameter, decompress(ion), more H.264/AVC compress for the second time and obtain second-compressed video library with the second quantization parameter; Described the first quantization parameter is less than the second quantization parameter.Unpressed video model sequence adopts the uncompressed video model sequence in step (S111), step (S112) is first done a H.264/AVC compression with the first quantization parameter QP1 to each yuv video sequence, then carry out decompress(ion), H.264/AVC, after decompress(ion), making secondary again with the second quantization parameter QP2 compresses, guarantee that the first quantization parameter QP1 is less than the second quantization parameter QP2, obtains second-compressed video library.
(S12) respectively the first compression video sequence in first compression video library and the second-compressed video sequence in second-compressed video library are carried out to entropy decoding, the summation about non-zero DCT coefficients after the I frame that obtains each first compression video sequence and each second-compressed video sequence quantizes and the DCT coefficient after inverse quantization; It specifically carries out each first compression video sequence entropy decoding and obtains summation about non-zero DCT coefficients after the I frame of each first compression video sequence quantizes and the DCT coefficient after inverse quantization, and each second-compressed video sequence is carried out to entropy decoding and obtain summation about non-zero DCT coefficients and the DCT coefficient after inverse quantization after the I frame of each second-compressed video sequence quantizes.
(S13) set of eigenvectors that the summation about non-zero DCT coefficients after the first compression video sequence that basis is obtained and the I frame of second-compressed video sequence quantize and the DCT coefficient after inverse quantization extract respectively first compression video sequence and second-compressed video sequence; Its concrete steps are:
(S131) coefficient threshold range and frequency threshold scope are set; Coefficient threshold range is set to [1,10], and 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 respectively each first compression video sequence and each second-compressed video sequence quantizes;
(S133) the summation about non-zero DCT coefficients medium frequency threshold range (0 after the I frame that extracts respectively each first compression video sequence and each second-compressed video sequence quantizes, 1) and (1,0) ac coefficient in, then the probability distribution of the ac coefficient absolute value in calculated rate threshold range in coefficient threshold range [1,10] respectively;
(S134) probability distribution of the ac coefficient absolute value in the summation about non-zero DCT coefficients after the I frame that calculates respectively each first compression video sequence and each second-compressed video sequence quantizes in all frequencies in coefficient threshold range [1,10];
(S135) the DCT coefficient histogram after the I frame inverse quantization of each first compression video sequence and each second-compressed video sequence is carried out respectively to fast fourier transform, obtain the energy spectral density of each first compression video sequence and each second-compressed video sequence;
(S136) extract respectively maximum point and the minimum point in the energy spectral density of each first compression video sequence and each second-compressed video sequence, choose an equal number maximum maximum point and minimum minimum point constitutive characteristic vector, as choose 3 maximum maximum points and 3 minimum minimum point constitutive characteristic vectors, obtain the set of eigenvectors of first compression video sequence and second-compressed video sequence; The probability distribution of calculating in step (S132)-(S134), each step obtains 10 in scope [1,10] frequency distribution value, therefore totally 30 dimensional features, add 3 maximum maximum points and 3 minimum minimum points, form the characteristic vectors of 36 dimensions of first compression video sequence, and the characteristic vectors of 36 dimensions of second-compressed video sequence, latter two characteristic vector constitutive characteristic vector set.
(S14) utilize grader to train and obtain classification based training model the set of eigenvectors of first compression video sequence and second-compressed video sequence; Grader can adopt SVMs.
(S2) utilize classification based training model to carry out the detection of second-compressed video to video to be detected, its specifically:
(S21) extract the characteristic vector of video to be measured, utilize classification based training model to predict classification, video to be measured is divided into two classes: first compression video and second-compressed video; Its concrete steps are:
(S211) video to be detected is carried out to entropy decoding, the summation about non-zero DCT coefficients after obtaining I frame and quantizing and the DCT coefficient after inverse quantization;
(S212) coefficient threshold range and frequency threshold scope are set; Coefficient threshold range is set to [1,10], and 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 that calculates video to be detected quantizes;
(S214) the summation about non-zero DCT coefficients medium frequency threshold range (0,1) after the I frame that extracts video to be detected quantizes and the ac coefficient in (1,0), the probability distribution of the ac coefficient absolute value in calculated rate threshold range in coefficient threshold range [1,10];
(S215) probability distribution of the ac coefficient absolute value in the summation about non-zero DCT coefficients after the I frame that calculates video to be detected quantizes in all frequencies in coefficient threshold range [1,10];
(S216) the DCT coefficient histogram after the I frame inverse quantization of video to be detected is carried out respectively to fast fourier transform, obtain the energy spectral density of video to be detected;
(S217) extract maximum point and the minimum point in the energy spectral density of video to be detected, choose equal number maximum maximum point and a minimum minimum point and form the characteristic vector of video to be detected, as choose the characteristic vector that 3 maximum maximum points and 3 minimum minimum points form video to be detected;
(S218) according to predicting classification in the characteristic vector input classification based training model of the video to be detected obtaining, video to be measured is divided into two classes: first compression video and second-compressed video.
Below in conjunction with specific embodiment, method of the present invention is described in detail, be to be noted that described embodiment is intended to be convenient to the understanding of the present invention, does not play restriction effect to the present invention.
First construct first compression video library, choose the yuv video sequence of 30 1665 frames and carry out first compression with X264 encoder, in order to increase the quantity of video library, 30 yuv video sequence Yi10Zheng Wei units are cut apart, each subsequence contains 10 frames.Each subsequence is compressed, and choosing the second quantization parameter QP2 is { 22,24,26,28,30}.
Next constructs second-compressed video library, first each yuv video subsequence is carried out to first compression with the first quantization parameter QP1, after decompress(ion), with the second quantization parameter QP2, carry out second-compressed again, wherein QP2 is greater than QP1, QP2 value is the same with the second quantization parameter QP2 that structure first compression video library is used, be 22,24,26,28,30, QP1 span is QP1=QP2-x, x=1,2,3,4,5.
The extraction of characteristic vector, each yuv video subsequence in first compression video library and second-compressed video library is carried out to entropy decoding, DCT coefficient after non-zero ac coefficient, DC coefficient and inverse quantization after the I frame that extracts each yuv video subsequence quantizes, the DCT coefficient after non-zero ac coefficient, DC coefficient and inverse quantization after quantizing according to the I frame of each yuv video subsequence of extracting is carried out respectively described step (S141)-(S146) the extract set of eigenvectors of first compression video sequence and second-compressed video sequence.
Structural classification device, utilizes support vector machines grader to train obtain two set of eigenvectors, obtains one and can judge whether video has passed through the classification based training model of second-compressed in the situation that of QP1<QP2.
Video to be detected is carried out to second-compressed detection, video to be detected is carried out to entropy decoding, DCT coefficient after non-zero ac coefficient, DC coefficient and inverse quantization after the I frame that extracts video to be detected quantizes, carry out described step (S131)-(S136) the extract characteristic vector of video to be detected, be input in classification based training model and train differentiation, with this, distinguish first compression video and second-compressed video.Experimental result is as shown in table 1 below:
In upper table, be to utilize the embodiment of the present invention to set up classification based training model 100 videos to be detected to be carried out to the testing result of secondary detection, from upper table, can know that the accuracy rate of utilizing the present invention to carry out second-compressed detection to video can reach more than 80%.As shown in Figure 3, the 1st curve represents the energy spectral density curve of first compression video sequence, the second quantization parameter using is QP2=26, other curves represent the energy spectral density curve of second-compressed video sequence, the first quantization parameter is respectively QP1=21,23,25, the second quantization parameter is QP2=26, as can be seen from Figure 3, and the analog value that the numerical value of the extreme point of the energy spectral density of second-compressed video sequence all obtains higher than first compression video sequence.Therefore can utilize extreme point in energy spectral density to distinguish the video of first compression and second-compressed.The present invention can effectively detect the video of second-compressed, is particularly useful for compressing for the second time rear video mass ratio and compresses for the first time the low-quality situation of rear video.
Claims (5)
1. based on a video second-compressed detection method H.264/AVC, it is characterized in that, comprise the steps:
(1) set up second-compressed visual classification training pattern, its specifically:
(11) unpressed video model sequence is carried out respectively to first compression and second-compressed and obtain first compression video library and second-compressed video library;
(12) respectively the first compression video sequence in first compression video library and the second-compressed video sequence in second-compressed video library are carried out to entropy decoding, the summation about non-zero DCT coefficients after the I frame that obtains each first compression video sequence and each second-compressed video sequence quantizes and the DCT coefficient after inverse quantization;
(13) set of eigenvectors that the summation about non-zero DCT coefficients after the first compression video sequence that basis is obtained and the I frame of second-compressed video sequence quantize and the DCT coefficient after inverse quantization extract respectively first compression video sequence and second-compressed video sequence;
(14) utilize grader to train and obtain classification based training model the set of eigenvectors of first compression video sequence and second-compressed video sequence;
(2) utilize classification based training model to carry out the detection of second-compressed video to video to be detected, its specifically:
(21) extract the characteristic vector of video to be measured, utilize classification based training model to predict classification, video to be measured is divided into two classes: first compression video and second-compressed video;
The concrete steps of described step (11) are:
(111) unpressed video model sequence is carried out to H.264/AVC compression acquisition first compression video library one time with the second quantization parameter;
(112) after H.264/AVC unpressed video model sequence is compressed for the first time with the first quantization parameter, decompress(ion), more H.264/AVC compress for the second time and obtain second-compressed video library with the second quantization parameter; Described the first quantization parameter is less than the second quantization parameter;
Described step (13) concrete steps are:
(131) coefficient threshold range and frequency threshold scope are set;
(132) probability distribution of DC coefficient absolute value in coefficient threshold range in the summation about non-zero DCT coefficients after the I frame that calculates each first compression 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 first compression video sequence and each second-compressed video sequence quantizes, the probability distribution of the ac coefficient absolute value in calculated rate threshold range in coefficient threshold range;
(134) probability distribution of the ac coefficient absolute value in the summation about non-zero DCT coefficients after the I frame that calculates each first compression video sequence and each second-compressed video sequence quantizes in all frequencies in coefficient threshold range;
(135) the DCT coefficient histogram after the I frame inverse quantization of each first compression video sequence and each second-compressed video sequence is carried out to fast fourier transform, obtain energy spectral density;
(136) extract maximum point and the minimum point in the energy spectral density of each first compression video sequence and each second-compressed video sequence, choose an equal number maximum maximum point and minimum minimum point, by the frequency distribution value constitutive characteristic vector obtaining in the value of choosing and step (132)-(134), obtain the set of eigenvectors of first compression video sequence and second-compressed video sequence.
2. according to claim 1 based on video second-compressed detection method H.264/AVC, it is characterized in that, in described step (21), extract the characteristic vector of video to be measured, utilize classification based training model to predict classification, video to be measured be divided into two classes: the concrete steps of first compression video and second-compressed video are:
(211) video to be detected is carried out to entropy decoding, the summation about non-zero DCT coefficients after obtaining I frame and quantizing and the DCT coefficient after inverse quantization;
(211) coefficient threshold range and frequency threshold scope are set;
(212) probability distribution of DC coefficient absolute value in coefficient threshold range in the summation about non-zero DCT coefficients after the I frame that calculates video to be detected quantizes;
(213) ac coefficient in the summation about non-zero DCT coefficients medium frequency threshold range after the I frame that extracts video to be detected quantizes, the probability distribution of the ac coefficient absolute value in calculated rate threshold range in coefficient threshold range;
(214) probability distribution of the ac coefficient absolute value in the summation about non-zero DCT coefficients after the I frame that calculates video to be detected quantizes in all frequencies in coefficient threshold range;
(215) the DCT coefficient histogram after the I frame inverse quantization of video to be detected is carried out to fast fourier transform, obtain energy spectral density;
(216) extract maximum point and the minimum point in the energy spectral density of video to be detected, choose the characteristic vector that an equal number maximum maximum point and minimum minimum point form video to be detected;
(217) according to predicting classification in the characteristic vector input classification based training model of the video to be detected obtaining, video to be measured is divided into two classes: first compression video and second-compressed video.
According to described in claim 1 to 2 any one based on video second-compressed detection method H.264/AVC, it is characterized in that, described coefficient threshold range is set to [1,10], described frequency threshold scope is set to (0,1) and (1,0).
According to described in claim 1 to 2 any one based on video second-compressed detection method H.264/AVC, it is characterized in that, described video model sequence is yuv video sequence.
According to described in claim 1 to 2 any one based on video second-compressed detection method H.264/AVC, it is characterized in that, described grader is 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 CN102547371A (en) | 2012-07-04 |
| CN102547371B true 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) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| 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 |
| CN105448299B (en) * | 2015-11-17 | 2019-04-05 | 中山大学 | A method of identifying digital audio AAC format codec |
| CN106331730B (en) * | 2016-08-22 | 2020-03-06 | 上海交通大学 | H.264 video same quantization factor double compression detection method |
Citations (4)
| 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 |
| 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 |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8260067B2 (en) * | 2008-04-18 | 2012-09-04 | New Jersey Institute Of Technology | Detection technique for digitally altered images |
-
2012
- 2012-02-27 CN CN201210045742.7A patent/CN102547371B/en not_active Expired - Fee Related
Patent Citations (4)
| 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 |
| 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 (6)
| Title |
|---|
| JPEG二次压缩的分析与检测;戴蒙;《中国图象图形学报》;20091231;第14卷(第12期);全文 * |
| 一种新的JPEG图像二次压缩检测算法;韩晓东;《计算机工程》;20100228;第36卷(第4期);第141页第1－2段,3.3节 * |
| 基于JPEG二次压缩的隐藏信息检测;戴蒙;《华东理工大学学报(自然科学版)》;20071031;第33卷(第5期);全文 * |
| 戴蒙.JPEG二次压缩的分析与检测.《中国图象图形学报》.2009,第14卷(第12期), |
| 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)》.2007,第33卷(第5期), |
| 韩晓东.一种新的JPEG图像二次压缩检测算法.《计算机工程》.2010,第36卷(第4期), |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102547371A (en) | 2012-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP4566498B2 (en) | How to describe motion activity in video | |
| Sun et al. | Exposing video forgeries by detecting MPEG double compression | |
| 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 | |
| CN101790097B (en) | Method for detecting multiple times of compression and coding of digital video | |
| Jiang et al. | Detection of HEVC double compression with the same coding parameters based on analysis of intra coding quality degradation process | |
| 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 | |
| Xu et al. | Detection of double MPEG-2 compression based on distributions of DCT coefficients | |
| KR101149522B1 (en) | Apparatus and method for detecting scene change | |
| CN103067713B (en) | Method and system of bitmap joint photographic experts group (JPEG) compression detection | |
| Jiang et al. | Detection of double compressed HEVC videos using GOP-based PU type statistics | |
| Bakas et al. | Mpeg double compression based intra-frame video forgery detection using cnn | |
| CN104853186A (en) | Improved video steganalysis method based on motion vector reply | |
| CN103020138A (en) | Method and device for video retrieval | |
| Bagdanov et al. | Adaptive video compression for video surveillance applications | |
| CN105915916B (en) | Video steganalysis method based on the estimation of motion vector distortion performance | |
| Mahfoudi et al. | Statistical H. 264 double compression detection method based on dct coefficients | |
| CN102595136A (en) | H.264/AVC (automatic volume control) video secondary compression detection method based on quantized coefficient statistical properties | |
| Xu et al. | Detection of video transcoding for digital forensics | |
| CN103034993A (en) | Digital video transcode detection method | |
| CN108366295A (en) | Visual classification feature extracting method, transcoding weight contracting detection method and storage medium | |
| Li et al. | Double HEVC compression detection with different bitrates based on co-occurrence matrix of PU types and DCT coefficients | |
| Xu et al. | Detection of transcoded HEVC videos based on in-loop filtering and PU partitioning analyses |
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 |