CN106375768A - Video steganalysis method based on intra prediction mode calibration (IPMC) - Google Patents
Video steganalysis method based on intra prediction mode calibration (IPMC) Download PDFInfo
- Publication number
- CN106375768A CN106375768A CN201510438280.9A CN201510438280A CN106375768A CN 106375768 A CN106375768 A CN 106375768A CN 201510438280 A CN201510438280 A CN 201510438280A CN 106375768 A CN106375768 A CN 106375768A
- Authority
- CN
- China
- Prior art keywords
- calibration
- ipm
- video
- satd
- pieces
- 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
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention relates to a video steganalysis method based on intra prediction mode calibration (IPMC). The method comprises steps: 1) IPMC is carried out on a to-be-detected video, the IPMC carries out block-by-block decompression and secondary compression with a 4*4 block as a unit, each IPM is traversed during the secondary compression process, and a corresponding sum of absolute transformed differences (SATD) is calculated; 2) according to data recorded during the calibration process, an IPMC feature set is calculated, wherein the IPMC feature set comprises two sub sets of an IPM transition probability feature set and an SATD transition distance feature set; and 3) the IPMC feature set is provided for a classifier for learning and classification, and steganalysis is further carried out. The existing video steganalysis algorithm based on the IPM can be effectively detected.
Description
Technical field
The invention belongs to the steganalysis field in information security, it is related to h.264/avc a kind of under video encoding standard be based on frame in
The video steganalysis method of predictive mode calibration.
Background technology
Present information steganography to reach secret communication by being embedded into secret information to seem in normal digital media carrier
Purpose.Steganography carrier widely, including text, audio frequency, image, video etc..The audio visual of steganography carrier and statistical property
All quite similar with non-steganography carrier it is difficult to by conventional method respectively.Video, with respect to text, audio frequency and image, has bigger
Data volume, wider application scenarios.This also means that video steganography carrier has bigger steganographic capacity and wider biography
Broadcast approach, just due to these advantages, video steganography is increasingly becoming the emphasis of research.
Information steganography technology species is various, is widely used, and the safe transmission for protecting important information provides favourable guarantee.But
This technology is not only used by multinational security department, military forces and intelligence agency, also can by hostile force information spy, probably
Fear molecule and cult member are using the propagation carrying out various information, plot, reaction speech etc. and communication, serious prestige
Politics and the economic security of country are coerced.Therefore, the Steganalysis for resisting steganography become in recent years also becomes domestic
The emphasis of outer research.The ultimate principle of steganalysis is the statistical property or the judgement of some condition codes according to digital media file data
Whether it has been embedded into secret information in carrier, or even attempted extracting the secret information containing in secret carrier.Find through investigation, at present
Steganography with text, image as carrier all compares early with the research starting of Steganalysis, and achievement in research is also quite varied.Mesh
The research in front domestic and international video steganography field is also more, is just continuously emerging outstanding achievement in research；But video steganography is divided
The correlational study achievement of analysis is then little, causes the unbalance situation of attacking and defending.Therefore, in order to effectively contain the indiscriminate of video steganography
With the Network Communicate Security problem leading to, need the correlational study to video Steganalysis for the booster injection badly.
Video steganography field, occurs in that the video steganographic algorithm based on intra prediction mode (ipm:intra prediction mode),
Such algorithm is generally basede on h.264/avc video encoding standard, that is, the intra prediction mode passing through to change macro block in i frame embeds
Secret information.For example: have scholar to propose to set certain mapping ruler first by 9 kinds of predictive modes in " 4x4 infra-frame prediction "
(Fig. 1) be mapped to binary number 0 or 1, under this mapping ruler, further according to secret information and associate(d) matrix coding, to i frame
The intra prediction mode of middle macro block is modified.In conjunction with the steganography coding techniques such as matrix coder, such steganographic algorithm can reach
Steganography safety well, also has higher steganographic capacity.For the steganographic algorithm based on intra prediction mode, there is scholar's base
Devise a kind of steganalysis method in Markov chain, but the supposed premise of the method is excessively preferable, the performance in practical application
Unsatisfactory, can not the video steganographic algorithm based on intra prediction mode for the effective detection.
Content of the invention
For solving the above problems, the present invention proposes a kind of video steganalysis method based on intra prediction mode calibration, Neng Gouyou
Effect ground detects the existing video steganographic algorithm based on intra prediction mode.
The present invention carries out video intra-frame prediction model calibration first, will video to be checked decompress and secondary with 4 × 4 pieces for unit block-by-block
Compression, and reselect infra-frame prediction during second-compressed；In a calibration process record related data be used for calculating frame in pre-
Survey model calibration (ipmc:intra prediction mode calibration) feature；Then ipmc feature is supplied to grader
Such as support vector machine (svm:support vector machine) are used for learning and classify, and then carry out steganalysis.
The overall technical architecture of the video steganalysis method based on intra prediction mode calibration of the present invention includes infra-frame prediction
Model calibration, ipmc feature calculation and svm support vector machine study with classification three big steps, main-process stream as shown in Fig. 2
Implementation steps include:
Step 1: carry out intra prediction mode calibration with 4 × 4 pieces for unit, be possibly used for embedded 4 × 4 in screening i frame first
Block, then by these 4 × 4 pieces of block-by-block decompression simultaneously second-compressed, travels through 9 kinds of predictive modes during second-compressed and calculates phase
The absolute transformed error answered and (satd:sum of absolute transformed differences).
Step 2: the related data according to record in step 1 extracts ipmc feature set, ipmc feature set comprises ipm transfer
Probability characteristics collection and two subsets of satd transfer distance feature set.Ipm transition probability feature set features the ipm after calibration
Deviate original optimum ipm probability, and satd transfer distance feature set feature the satd after calibration deviate original value away from
From.
Step 3: learnt using grader such as svm support vector machine and classify.During using svm support vector machine,
Kernel function adopts gaussian kernel function.If inputting as training sample video, ipmc feature is used for training grader template；If
Input as video to be checked, then mate corresponding grader template first, then be used for classifying to judge that video is by ipmc feature
No for steganography carrier.
The video steganography method of the present invention has the beneficial effect that to correlative technology field:
1. the present invention, with respect to existing similar steganalysis method, detects that performance has and is quite obviously improved.Currently for being based on
The steganalysis method achievement of infra-frame prediction mould steganography is less, and existing method supposed premise is excessively preferable, the table in practical application
Existing unsatisfactory.Calibration program in the present invention can accurately portray the distortion that steganography causes to video, extracts feature to frame in
The non-dominance of predictive mode is very sensitive.Even if video to be checked embed rate very low when, detection performance is still stable.
2. the present invention solves the problems, such as that in conventional video calibration scheme, second-compressed parameter is difficult to be adapted to.According to investigation, existing
Generally whole video is decompressed and second-compressed as an entirety based on the video steganalysis method of calibration, be usually present
Second-compressed parameter is difficult to be adapted to the problem leading to detect that performance drastically declines with compression first.And calibration program in the present invention
In, decompression is carried out with 4 × 4 pieces for unit with second-compressed, and all second-compressed parameters all can be from original input video
Obtain, not there is a problem of being difficult to be adapted to.
3. the technical scheme that the present invention is calibrated in units of block has good flexible expansion.The present invention selects with block as list
Position carries out calibrating the concordance that when ensure that second-compressed well, macro block divides, and so that the adaptation issues of second-compressed parameter is obtained
To solve.Based on this thinking, different calibration programs can be formulated to detect different steganalysis algorithms, for example, with macro block be
Motion vector calibration program of unit etc. is it is seen that the present invention has good flexible expansion.
Brief description
Fig. 1 is h.264/avc 9 kinds of intra prediction mode schematic diagrams in video encoding standard.
Fig. 2 is the video steganalysis method flow chart based on intra prediction mode calibration that the present invention provides.
Fig. 3 is the flow chart of the intra prediction mode calibration of the present invention.
Fig. 4 is the position relationship schematic diagram of current block and adjacent block during compressed encoding.
Fig. 5 is the fisrt feature collection scattergram of the present invention.
Fig. 6 is the second feature collection scattergram of the present invention.
Specific embodiment
Below in conjunction with accompanying drawing 3-6, the specific embodiment of the present invention is conducted further description.
Provided in an embodiment of the present invention intra prediction mode school is included based on the video steganalysis method of intra prediction mode calibration
Standard, ipmc feature calculation and three big steps of support vector machine study and classification, concrete operations flow process is as follows:
Step 1: carry out intra prediction mode calibration with 4 × 4 pieces for unit.It is possibly used for embedded 4 × 4 first in screening i frame
Block, then by these 4 × 4 pieces of block-by-block decompression simultaneously second-compressed, travels through 9 kinds of predictive modes during second-compressed and calculates phase
The absolute transformed error answered and (satd:sum of absolute transformed differences) (kim, j., jeong, j.:fast
intra mode decision algorithm using the sum of absolute transformed differences.in:
proceedings of 2011 international conference on digital image computing:techniques and
Applications, dicta 2011, pp.655-659.ieee (2011)), satd is defined as follows:
Wherein, a represents present encoding block, and s presentation code block compresses preceding pixel value set, srecRepresent the pixel after decompression reconstruction
Value set, (x, y) is pixel coordinate, and h (*) is Ha Deman transforming function transformation function.The enforcement of step 1 is specifically subdivided into following four sub-steps
(with reference to Fig. 3):
1) calibration will be carried out with 4 × 4 pieces in i frame for least unit, takes out i frame, and take out i frame from input video code stream
One of be likely to be used for steganography embed 4 × 4 pieces.In h.264 compressed encoding, in order to save memory space further,
Position relationship as shown in Figure 4, if the predictive mode of 4 × 4 pieces of c of present encoding is equal to block a and left block b prediction mould adjacent and above
During the smaller value of formula, encoder can arrange marker bit pre=1, now need not store the predictive mode of current block c again.Therefore,
All steganographic algorithms based on intra prediction mode are embedded all without select pre=1 4 × 4 pieces.Therefore in the present invention, carry out school
Also 4 × 4 pieces of all pre=1 will be skipped on time, and all think for remaining 4 × 4 pieces and be likely to be used for steganography.
2) original predictive mode and quantization parameter (qp:quantization parameter) in current 4 × 4 pieces of storage, then will
Current 4 × 4 pieces of solutions are depressed into spatial domain, obtain sets of pixel values.This step is carried out according to h.264 encoder normal process, carefully
Section repeats no more.
3) current 4 × 4 pieces of second-compressed, travel through 9 kinds of predictive modes and calculate corresponding satd, then in compression process
Obtain current 4 × 4 pieces of ipm-satd calibration set, required reference pixel value all can be from the adjacent block having decompressed before
Obtain in data, quantization parameter qp directly adopts step 2) the middle original qp value storing.
4) repeat sub-step 1) to 3) until there is no available 4 × 4 pieces.
Step 2: the related data according to record in step 1 extracts ipmc feature set, ipmc feature set comprises ipm transfer
Probability characteristics collection and two subsets of satd transfer distance feature set, calculating process includes following two sub-steps:
1) ipm transition probability feature set is calculated according to the intra prediction mode set that each participates in before and after 4 × 4 pieces of calibrations calibrated
(fisrt feature collection), characteristic dimension is 9.This feature set features the probability that the ipm after calibration deviates original optimum ipm,
Computing formula is as follows:
Wherein, x ∈ [1,9] is the feature sequence number of fisrt feature collection, and k is current frame number, lkIt is that participate in calibrating 4 × 4 pieces are total
Number.il∈ [1,9] is original ipm value,It is x-th ipm during current 4 × 4 pieces of ipm-satd calibration is gathered,
In formula (2):
Fig. 5 contrast shows the fisrt feature collection distribution situation of (0.2mb/s and 1mb/s) steganography and non-steganography video under different code checks,
In experiment we use a kind of video steganographic algorithm based on intra prediction mode that scholar bouchama proposes (bouchama, s.,
hamami,l.,aliane,h.:h.264/avc data hiding based on intra prediction modes for real-time
application.in:proceedings of the world congress on engineering and computer science,vol.1,
Pp.655-658 (2012)) make steganography video sample.As can be seen that for non-steganography video, the probability of the overwhelming majority falls
In first feature, this shows that most of ipm is consistent with original ipm after calibration.But for steganography video, bright
Show more probability to be distributed in other features that is to say, that the ipm after calibration has higher probability to deviate original ipm.
2) satd transfer distance feature set is calculated with set according to the absolute error that each participates in before and after 4 × 4 pieces of calibrations calibrated
(second feature collection), characteristic dimension is 4.This feature features the distance that the satd after calibration deviates original value.Satd's
Deviateing is probably caused by compression artefacts or ipm steganography, but the satd that compression artefacts cause deviates very little, especially code
Almost close to 0 when rate is sufficiently high, and ipm steganography cause satd deviate then will be relatively high many.The calculating of second feature collection is public
Formula is as follows:
Wherein, y ∈ [Isosorbide-5-Nitrae] is the feature sequence number of second feature collection, dlRepresent original satd,Represent the optimum ipm after calibration
Corresponding satd value, in formula (4):
Second feature collection actually depict discrete probability distribution on four intervals for the satd deviation distance.Using discrete probabilistic
Distribution can effectively reduce characteristic dimension.In above formula, β is interval division parameter.In general, the code of β and video
Rate is negatively correlated, and table 1 gives the reference value of β under different code checks.
The reference value of β under the different code check of table 1.
Fig. 6 contrast shows the second feature collection distribution feelings of (0.2mb/s and 1mb/s) steganography and non-steganography video under different code checks
Condition.Similar to fisrt feature collection, for non-steganography video, most probability fall in first feature, and this shows satd
Deviation distance is generally less than normal.And for steganography video, satd deviation distance substantially increases, more probability fall in other
In feature.
Above sub-step 1) and 2) in calculated two feature sets merge and obtain final product final ipmc feature set, feature lump
Dimension is tieed up for 9+4=13.
Step 3: learnt using svm support vector machine and classify, kernel function adopts gaussian kernel function.If inputting as training
Sample video, then be used for training grader template by the ipmc feature of extraction.With frame group (typically setting one group of 8 frame) as unit
Extract feature, the code check with group training sample should be identical or as close as possible；If inputting as video to be checked, first according to treating
Whether inspection video code rate mates corresponding grader template, then be used for classifying to judge video as hidden by the ipmc feature of extraction
Write carrier.Video code rate is the principal element of impact detection performance, therefore either learns and video code rate all should be made by classification
For important referential data.Svm support vector machine are prior art, and it will not go into details for this specification.
Specific embodiment described herein is only the explanation for example to general principles of the present invention.The technical field of the invention
Technical staff described specific embodiment can be made with various modification or supplement or do not substituted, but not using similar mode
The essential concept of the present invention can be deviateed or surmount scope defined in appended claims.
Claims (7)
1. a kind of video steganalysis method based on intra prediction mode calibration, its step includes:
1) video to be checked is carried out intra prediction mode calibration, described intra prediction mode is calibrated with 4 × 4 pieces for unit block-by-block solution
Press and carry out second-compressed, travel through various intra prediction mode ipm calculating during second-compressed and definitely become accordingly
Change error and satd；
2) data according to record in calibration process calculates intra prediction mode alignment features collection, described intra prediction mode calibration
Feature set comprises ipm transition probability feature set and two subsets of satd transfer distance feature set, and described ipm transfer is general
Rate feature set is the probability of the ipm deviation original optimum ipm after calibration, and described satd transfer distance feature set is calibration
Satd afterwards deviates the distance of original value；
3) described intra prediction mode alignment features collection is supplied to grader to be learnt and classify, and then carries out steganalysis.
2. the method for claim 1 is it is characterised in that step 1) includes following sub-step:
1-1) take out i frame from input video code stream, and take out one of i frame and be likely to be used for embedded 4 × 4 pieces of steganography；
1-2) store original predictive mode and quantization parameter in current 4 × 4 pieces, then current 4 × 4 pieces of solutions be depressed into spatial domain,
Obtain sets of pixel values；
1-3) current 4 × 4 pieces of second-compressed, travel through 9 kinds of predictive modes and calculate corresponding satd, so in compression process
Obtain current 4 × 4 pieces of ipm-satd calibration set afterwards, required reference pixel value is from the adjacent block data having decompressed
Middle acquisition, quantization parameter directly adopts step 1-2) the middle original quantisation coefficient value storing；
1-4) repeat sub-step 1-1) to 1-3) until there is no available 4 × 4 pieces.
3. the method for claim 1 it is characterised in that: step 2) according to each participate in calibration 4 × 4 pieces calibration before and after
Intra prediction mode set calculate described ipm transition probability feature set, its computing formula is as follows:
Wherein, x ∈ [1,9] is the feature sequence number of fisrt feature collection, and k is current frame number, lkIt is participate in calibration 4 × 4 pieces
Sum, il∈ [1,9] is original ipm value,It is xth during current 4 × 4 pieces of ipm-satd calibration is gathered
Individual ipm, in above-mentioned formula:
4. method as claimed in claim 3 it is characterised in that: step 2) according to each participate in calibration 4 × 4 pieces calibration before and after
Absolute error and set calculate satd transfer distance feature set, its computing formula is as follows:
Wherein, y ∈ [Isosorbide-5-Nitrae] is the feature sequence number of second feature collection, dlRepresent original satd,Represent the optimum after calibration
Ipm corresponding satd value, in above-mentioned formula:
In above formula, β is interval division parameter.
5. the method for claim 1 it is characterised in that: grader described in step 3) be svm support vector machine, its core
Function adopts gaussian kernel function.
6. the method as described in claim 1 or 5 it is characterised in that: in step 3), if inputting as training sample video, will
Ipmc feature is used for training grader template；If inputting as video to be checked, mate corresponding grader template first, so
Afterwards ipmc feature is used for classification to judge video whether as steganography carrier.
7. method as claimed in claim 6 it is characterised in that: step 3) is according to corresponding point of the video code rate of video to be checked coupling
Class device template.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510438280.9A CN106375768B (en) | 2015-07-23 | 2015-07-23 | Video steganalysis method based on intra prediction mode calibration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510438280.9A CN106375768B (en) | 2015-07-23 | 2015-07-23 | Video steganalysis method based on intra prediction mode calibration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106375768A true CN106375768A (en) | 2017-02-01 |
| CN106375768B CN106375768B (en) | 2019-05-17 |
Family
ID=57881000
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510438280.9A Expired - Fee Related CN106375768B (en) | 2015-07-23 | 2015-07-23 | Video steganalysis method based on intra prediction mode calibration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106375768B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109711121A (en) * | 2018-12-27 | 2019-05-03 | 清华大学 | Text steganography method and device based on Markov model and Huffman encoding |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101500161A (en) * | 2008-01-31 | 2009-08-05 | 华为技术有限公司 | Inter-frame prediction method and apparatus based on adaptive block transformation |
| US20110019907A1 (en) * | 2006-01-13 | 2011-01-27 | New Jersey Institute Of Technology | Method for identifying marked images using statistical moments based at least in part on a jpeg array |
| CN103765892A (en) * | 2011-06-28 | 2014-04-30 | 三星电子株式会社 | Method and apparatus for coding video and method and apparatus for decoding video, accompanied with intra prediction |
| CN104125467A (en) * | 2014-08-01 | 2014-10-29 | 郑州师范学院 | Embedding and extracting methods for video steganography information |
| CN104602005A (en) * | 2010-08-17 | 2015-05-06 | M&K控股株式会社 | Method for decoding intra-predictions |
-
2015
- 2015-07-23 CN CN201510438280.9A patent/CN106375768B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110019907A1 (en) * | 2006-01-13 | 2011-01-27 | New Jersey Institute Of Technology | Method for identifying marked images using statistical moments based at least in part on a jpeg array |
| CN101500161A (en) * | 2008-01-31 | 2009-08-05 | 华为技术有限公司 | Inter-frame prediction method and apparatus based on adaptive block transformation |
| CN104602005A (en) * | 2010-08-17 | 2015-05-06 | M&K控股株式会社 | Method for decoding intra-predictions |
| CN103765892A (en) * | 2011-06-28 | 2014-04-30 | 三星电子株式会社 | Method and apparatus for coding video and method and apparatus for decoding video, accompanied with intra prediction |
| CN104125467A (en) * | 2014-08-01 | 2014-10-29 | 郑州师范学院 | Embedding and extracting methods for video steganography information |
Non-Patent Citations (1)
| Title |
|---|
| 孔维国，王宏霞，王科人，刘正辉: "基于转移概率矩阵的H.264/AVC视频帧内预测模式信息隐藏检测算法", 《四川大学学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109711121A (en) * | 2018-12-27 | 2019-05-03 | 清华大学 | Text steganography method and device based on Markov model and Huffman encoding |
| CN109711121B (en) * | 2018-12-27 | 2021-03-12 | 清华大学 | Text steganography method and device based on Markov model and Huffman coding |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106375768B (en) | 2019-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102917227B (en) | Compressive sensing-based adaptive video information hiding method | |
| CN103034853B (en) | A kind of jpeg image general steganalysis method | |
| Nie et al. | Defining Embedding Distortion for Intra Prediction Mode-Based Video Steganography. | |
| CN103345767B (en) | A kind of JPEG image steganography method of high security | |
| CN104661037B (en) | The detection method and system that compression image quantization table is distorted | |
| CN110913092B (en) | Reversible information hiding method for encrypted image | |
| CN104284190B (en) | Compressed image steganography encoding method based on AMBTC high-low mean value optimization | |
| CN110162986B (en) | Reversible information hiding method based on adjacent pixel prediction model | |
| CN105374054A (en) | Hyperspectral image compression method based on spatial spectrum characteristics | |
| CN102857831B (en) | H.264 video integrality authentication method | |
| CN102156955A (en) | Robust reversible watermark embedding and extracting method based on histogram neighborhood | |
| CN105631469A (en) | Bird image recognition method by multilayer sparse coding features | |
| Gan et al. | Video object forgery detection algorithm based on VGG-11 convolutional neural network | |
| Kumar et al. | Near lossless image compression using parallel fractal texture identification | |
| CN106231356A (en) | The treating method and apparatus of video | |
| Xie et al. | Bag-of-words feature representation for blind image quality assessment with local quantized pattern | |
| Niu et al. | Machine learning-based framework for saliency detection in distorted images | |
| CN103903214B (en) | Method for assessing DCT-domain image steganography capacity based on MCUU model | |
| Li et al. | Steganography of steganographic networks | |
| CN106375768A (en) | Video steganalysis method based on intra prediction mode calibration (IPMC) | |
| CN100535926C (en) | Data processing method and apparatus, image processing method and apparatus, image sorting method and apparatus, and storage medium | |
| CN106101713B (en) | A kind of video steganalysis method based on the optimal calibration of window | |
| CN107454408A (en) | A kind of method of Image Coding code check dynamic adjustment | |
| CN115955376A (en) | Modulation identification method based on convolutional neural network and DAE _ Transformer | |
| CN106530365A (en) | Self-adaptive compressed sensing reconstruction method based on image information content difference |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190517 Termination date: 20190723 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |