CN112637605A - Video steganalysis method and device based on analysis of CAVLC code words and number of nonzero DCT coefficients - Google Patents
Video steganalysis method and device based on analysis of CAVLC code words and number of nonzero DCT coefficients Download PDFInfo
- Publication number
- CN112637605A CN112637605A CN202011253138.4A CN202011253138A CN112637605A CN 112637605 A CN112637605 A CN 112637605A CN 202011253138 A CN202011253138 A CN 202011253138A CN 112637605 A CN112637605 A CN 112637605A
- Authority
- CN
- China
- Prior art keywords
- video
- dct coefficients
- steganalysis
- sub
- cavlc
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/625—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using discrete cosine transform [DCT]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/46—Embedding additional information in the video signal during the compression process
- H04N19/467—Embedding additional information in the video signal during the compression process characterised by the embedded information being invisible, e.g. watermarking
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
Abstract
The invention relates to a video steganalysis method and a device based on analysis of CAVLC code words and non-zero DCT coefficient numbers. The method comprises the following steps: dividing a compressed video to be detected into a plurality of frame groups; for each frame group, obtaining video steganalysis characteristics by analyzing the correlation between CAVLC entropy coding code words and the number of nonzero DCT coefficients of adjacent 4 multiplied by 4 brightness macro blocks; and performing steganography classification judgment on each frame group in the compressed video to be detected by using the obtained video steganography analysis characteristics and adopting a classifier. The invention extracts four types of sub-features and finally obtains the 635-dimensional steganalysis feature. The method can effectively detect the existing video steganography method of the DCT coefficient domain, can effectively relieve the carrier source mismatch phenomenon in steganography analysis to a certain extent, has a wide application range, and can effectively analyze the video steganography of the DCT coefficient domain based on different video coding standards.
Description
Technical Field
The invention relates to the field of Video Steganalysis (Video Steganalysis), belongs to the information hiding sub-field in the technical field of information security, and particularly relates to a method and a device for analyzing Video steganography in a DCT coefficient field based on analysis of CAVLC code words and the number of non-zero DCT coefficients.
Background
The purpose of steganography is to obscure the fact and information content that secret information is transmitted. In contrast, the purpose of steganalysis is to determine whether there is an embedded secret message in a multimedia file. In recent years, with the rapid development of streaming media and internet video services, digital video is gradually replacing images as the most influential media. Therefore, video steganography and steganalysis become the current research focus.
Commonly used embedding elements include Intra Prediction Mode (Intra Prediction Mode), Inter Prediction Mode (Inter Prediction Mode), Motion Vector (Motion Vector), Quantization Parameters (Quantization Parameters), and DCT Coefficients (Discrete Cosine Transform Coefficients). In the common embedded elements, the DCT coefficient accounts for the highest proportion of the H.264/AVC video code stream. Therefore, the video steganography method based on DCT coefficients generally has a large embedding capacity. Therefore, video steganography methods based on DCT coefficients have recently received much attention from scholars in the field.
Video steganography methods based on DCT coefficients can be divided into two categories according to the steganography principle: CDE (Compression Domain Embedding) and JCE (Joint Compression Embedding, combined Compression Embedding) (references: Y. Cao, Y. Wang, X ZHao, M Zhu, and Z Xu, "Cover Block deconfiguration for Content-Adaptive H.264 Steganographic," in Proc.6th ACM Workshop Inf.Hiding Multimedia Security, Austria, June.2018, pp.23-30). In the first type of video steganography method based on DCT coefficients, entropy decoding a video code stream to obtain DCT coefficients, modifying the coefficients, and directly entropy coding the modified coefficients into the video code stream again. The method has small calculation amount and can well meet the real-time processing requirement. However, such methods have the problem of "distortion drift", that is, if the coefficients are modified at will, the errors introduced by the coefficient modification during decoding will accumulate continuously, thereby affecting the visual quality of the subsequent video macro block. Therefore, to solve the "Distortion Drift" problem, Ma, etc. compensate for Distortion by coupling coefficient pairs (each consisting of two DCT coefficients), and when modifying one coefficient of the coupling coefficient pair, compensate for the other coefficient (references: X.Ma, Z.Li, H.Tu, and B.Zhang, "A Data high Algorithm for H.264/AVC Video Streams with Intra-Frame Distortion drive," IEEE Transactions on Circuits Systems and s for Video Technology, vol.20, No.5, pp.1320-1330, Oct.2010.). In the second category of video steganography methods based on DCT coefficients, the modification of the coefficients occurs during the video compression process. This type of method, while requiring more computational overhead, can completely avoid the problem of "distortion drift". Cao et al propose a DCT coefficient steganography method based on inter-block decoupling, successfully suppressing the 'cost drift'. "cost drift", i.e. modification of the current element during the embedding process, changes the embedding cost of the subsequent element (ref: Y.Cao, Y.Wang, X ZHao, M Zhu, and Z Xu, "Cover Block cloning for Content-Adaptive H.264Steganograph," in Proc.6th ACM Workshop Inf.Hiding Multimedia Security, Austria, June.2018, pp.23-30). Meanwhile, the method applies STC (synchronous Trellis Codes) embedded messages, and the whole embedding influence is reduced to the minimum.
Through patent inquiry, the related invention patent application cases existing in the field are as follows:
patent application No. CN201710447336.6, "a video steganalysis method for detecting steganalysis based on DCT coefficients," discloses a video steganalysis method in DCT coefficient domain. According to the method, a spatial domain histogram is obtained by calculating a spatial domain characteristic set by using a DCT (discrete cosine transformation) kernel and an embedding cost according to the influence of steganography operation based on DCT coefficient domain steganography on a video pixel spatial domain and time domain correlation. And inter-frame similar macro blocks are connected through the motion vectors, space domain slicing is constructed, and a time domain characteristic set is calculated to obtain a time domain histogram. And combining the spatial histogram and the time domain histogram into a final 1440-dimensional steganalysis feature set. Because the video steganography method based on the DCT coefficient domain belonging to the JCE has small influence on the spatial domain and the time domain of the video, the steganography analysis capability of the method on the DCT coefficient domain belonging to the JCE is weak. And the calculation of the method involves operations such as convolution and the like, and the calculation amount is large. The invention does not relate to the calculation of the spatial domain and the temporal domain correlation of the video pixels, and only concerns the DCT coefficient and the CAVLC entropy coding code word corresponding to the DCT coefficient, so the method is obviously different from the design thought and the specific implementation mode of the invention.
Disclosure of Invention
The invention provides a video steganalysis method and a device with high steganalysis capacity and simple calculation for a DCT coefficient domain, and aims to judge whether DCT coefficients in a video are modified or not by analyzing the correlation between CAVLC entropy coding code words and the number of nonzero DCT coefficients of adjacent 4 multiplied by 4 brightness macro blocks.
Compared with other DCT coefficient domain video steganalysis methods, the method provided by the invention directly analyzes the correlation between CAVLC entropy Coding code words of DCT coefficients and the number of nonzero DCT coefficients of adjacent 4 x 4 brightness macro blocks, designs two characteristics to describe CAVLC (Context Adaptive Variable Length Coding, Context-based Variable Length Coding) entropy Coding code words and two characteristics to describe the correlation between the number of nonzero DCT coefficients of adjacent 4 x 4 brightness macro blocks 635, and finally designs the dimensional DCT coefficient domain video steganalysis characteristics.
According to research, the currently existing video steganalysis method of the DCT coefficient domain has the following two limitations: firstly, the existing analysis method carries out a large amount of convolution operations on the space domain and the time domain of a video frame, and the calculated amount is large. Secondly, the existing analysis method only considers the spatial domain and time domain correlation of the video frame, the DCT coefficient video steganography method belonging to JCE has small influence on the spatial domain and the time domain of the video frame, and STC embedding information is adopted to minimize the whole embedding influence. Therefore, the existing analysis method has lower accuracy in analyzing the video steganography method of DCT coefficients belonging to JCE.
The technical scheme adopted by the invention is as follows:
a video steganalysis method based on analyzing CAVLC code words and non-zero DCT coefficient numbers comprises the following steps:
dividing a compressed video to be detected into a plurality of frame groups;
for each frame group, obtaining video steganalysis characteristics by analyzing the correlation between CAVLC entropy coding code words and the number of nonzero DCT coefficients of adjacent 4 multiplied by 4 brightness macro blocks;
and performing steganography classification judgment on each frame group in the compressed video to be detected by using the obtained video steganography analysis characteristics and adopting a classifier.
Further, the above method comprises the following steps (as shown in fig. 1):
1) frame group division: the method comprises the steps of dividing a compressed video to be detected into a plurality of frame groups, wherein each frame group consists of continuous video frames, and the video frames are not overlapped and have the same number.
2) For a certain frame group F comprising N4 x 4 luminance macroblocksgAnd executing steps 3) to 6) to extract the steganalysis characteristics.
3) Pretreatment: for frame group FgEach 4 x 4 luminance macroblock in the block obtains its corresponding CAVLC codeword Bn(n∈[1,N]). If the video is entropy-coded in other modes (namely, the video adopting non-CAVLC entropy coding standard), decoding to obtain the corresponding DCT coefficient, and then CAVLC coding to obtain the corresponding CAVLC code word Bn。
4) Number analysis of non-zero DCT coefficients: decoding CAVLC codeword BnRecording the relative position (x, y) of the macroblock in the video frame (x, y) ((
H is the width of the video frame and W is the length of the video frame), and the number of corresponding non-zero DCT coefficients, the correlation of the number of non-zero DCT coefficients of adjacent macroblocks is calculated.
5) Code word analysis: traversal frame set FgAll CAVLC code words B innAnd calculating the proportion of '1' in the code words under different numbers of the non-zero DCT coefficients and the proportion of '1' in each position of all the code words.
6) And (3) feature calculation, namely extraction: according to the calculation results of the step 4) and the step 5), frame group F is setgAnd extracting preset steganalysis characteristics.
7) Steganalysis: and respectively carrying out steganography classification judgment on each frame group in the video to be tested by adopting a classifier based on preset steganography analysis characteristics.
Further, as for step 6), the 635-dimensional high-performance steganalysis feature provided by the present invention mainly comprises steganalysis features based on codeword analysis and steganalysis features based on non-zero DCT coefficient number correlation analysis, which will be described in detail below.
The invention firstly counts the proportion of '1' in the code word, and selects different code tables for coding according to the number of nonzero DCT coefficients of the macro block during CAVLC entropy coding. Therefore, when the code word is analyzed, the proportion of '1' in the code word under different numbers of the nonzero DCT coefficients is counted. And secondly, counting the proportion of the occurrence of '1' at each position in the code word. Meanwhile, the correlation among the number of the nonzero DCT coefficients between the blocks is considered during CAVLC coding, and the correlation among the number of the nonzero DCT coefficients between the blocks is described by calculating the conditional probability of the nonzero DCT between the upper block and the lower block and between the left block and the right block.
For a certain group of frames F in a given compressed videogAccording to the following stepsgAnd extracting four types of sub-features and finally forming to obtain 635-dimensional steganalysis features.
a) Pretreatment: for FgEach 4 × 4 luma macroblock in (B) gets its corresponding CAVLC codeword Bn(n∈[1,N]) Decoding CAVLC code word to obtain the number of non-zero DCT coefficients
Where (x, y) is the relative position of the block in the video frame (x, y)
H is the width of the video frame and W is the length of the video frame). Statistics BnNumber of middle "1
And the number of "0
And record BnCode word at each position in
If B isnIs automatically in B when the length of (2) is less than 40nThen zero padding is carried out; and if the value is more than 40, the operation is cut off.
b) Type 1 sub-feature extraction (based on codeword analysis): the proportion of "1" in the codeword, for a given number k of non-zero DCT coefficients per dimension of the type 1 sub-feature, is defined as
c) Type 2 sub-feature extraction (based on codeword analysis): each dimension of the type 2 sub-feature corresponds to the proportion of '1' at the k-th position in the code word, and is defined as
Wherein the function
The same applies below.
d) Type 3 sub-feature extraction (based on correlation between the number of non-zero DCT coefficients between blocks): the type 3 sub-feature is related to the correlation between the number of DCT coefficients of the upper and lower adjacent blocks, and the probability that the number of DCT coefficients of the lower adjacent block is equal to j given the number of DCT coefficients of the upper block is equal to i in each dimension is defined as
e) Type 4 sub-feature extraction (based on correlation between the number of non-zero DCT coefficients between blocks): the probability that the number of non-zero DCT coefficients of the adjacent right block is equal to j is defined as the probability that the number of non-zero DCT coefficients of the left block is equal to i when the type 4 sub-feature is related to the correlation of the number of non-zero DCT coefficients of the left and right adjacent blocks, and the number of non-zero DCT coefficients of each dimension corresponding to the given left block is equal to i
f) And (4) final feature combination: the final 635-dimensional steganalysis feature F (k) is obtained by combining the above 4 types of sub-features and is defined as
Based on the same inventive concept, the invention also provides a video steganalysis device based on analyzing CAVLC code words and non-zero DCT coefficients by adopting the method, which comprises the following steps:
the frame group dividing module is used for dividing the compressed video to be detected into a plurality of frame groups;
the characteristic extraction module is used for analyzing the correlation between CAVLC entropy coding code words and the number of nonzero DCT coefficients of adjacent 4 multiplied by 4 brightness macro blocks to obtain video steganalysis characteristics for each frame group;
and the steganography analysis module is used for performing steganography classification judgment on each frame group in the compressed video to be detected by using the obtained video steganography analysis characteristics and adopting a classifier.
The video steganalysis method of the invention has the following beneficial effects in the relevant technical field:
1) the video steganography method of the existing DCT coefficient domain can be effectively detected. In the video encoding process, the DCT coefficients are reordered and then entropy encoded. Since entropy coding is a lossless compression, any modification on the DCT coefficients will affect the result of entropy coding the codeword. Furthermore, the DCT coefficients of neighboring blocks have a correlation, and modifying the DCT coefficients affects this correlation. The invention constructs the relativity of the characteristic description entropy coding code word and the DCT coefficient number of the adjacent block, and utilizes the influence of the video steganography of the DCT coefficient field on the relativity of the nonzero DCT coefficient number of the entropy coding code word and the adjacent block, so the invention has ideal detection effect on the current video steganography method of the DCT coefficient field.
2) The carrier source mismatch phenomenon in steganalysis can be effectively relieved to a certain extent. The carrier Source Mismatch (Cover Source Mismatch) phenomenon in steganalysis refers to: when a steganalysis detector trained on one carrier source is used to analyze samples from different carrier sources, the difference between the two carrier sources will have a large negative impact on steganalysis accuracy. The carrier source mismatch phenomenon is ubiquitous in real network environments, and is the biggest obstacle to practical application of steganalysis. Because the invention describes the correlation of the entropy coding code word and the number of the nonzero DCT coefficients of the adjacent blocks by the structural characteristics, the invention effectively reflects the influence of video steganography in the DCT coefficient field on the correlation of the entropy coding code word and the number of the nonzero DCT coefficients of the adjacent blocks, and improves the steganography analysis performance and the stability thereof, so that the invention can effectively relieve the phenomenon of carrier source mismatch to a certain extent, such as: the steganography analysis detector obtained by training on the basis of extracting steganography analysis characteristics from samples with the same size, the same code rate and the same embedding rate by adopting the method has better analysis and detection effects on steganography videos with different sizes, different code rates and different embedding rates.
3) The method is widely applicable to different video coding standards. The invention mainly describes the correlation of the number of the nonzero DCT coefficients of the entropy coding code words and the adjacent blocks by constructing the characteristic, realizes that the video adopting the non-CAVLC entropy coding standard does not depend on a specific video coding standard, decodes to obtain the DCT coefficients of the video, and then carries out CAVLC entropy coding to carry out subsequent characteristic extraction operation. Therefore, the method has a wide application range, and can effectively analyze the video steganography of the DCT coefficient domain based on different video coding standards.
Drawings
FIG. 1 is a schematic illustration of video steganalysis employing the present invention;
fig. 2 is a flow chart of video steganalysis employing the present invention.
Detailed Description
The invention will now be further described by way of specific examples in conjunction with figure 2.
The flow of the video steganalysis method based on analyzing CAVLC codewords and non-zero DCT coefficients provided in this embodiment is shown in fig. 2, and the specific operation details are as follows:
1) frame group division: the method comprises the steps of dividing a compressed video to be detected into a plurality of frame groups, wherein each frame group consists of continuous video frames, and the video frames are not overlapped and have the same number.
2) For a certain frame group F comprising N4 x 4 luminance macroblocksgAnd executing steps 3) to 6) to extract the steganalysis characteristics.
3) Pretreatment: for frame group FgEach 4 x 4 luminance macroblock in the block obtains its corresponding CAVLC codeword Bn(n∈[1,N]). If the video is entropy-coded in other types, decoding to obtain the corresponding DCT coefficient, and then CAVLC coding to obtain the corresponding CAVLC code word Bn. Decoding CAVLC code word to obtain the number of DCT coefficients
Where (x, y) is the relative position of the block in the video frame (x, y)
H is the width of the video frame and W is the length of the video frame). Statistics BnNumber of middle "1
And the number of "0
And record BnCode word at each position in
If B isnIs automatically in B when the length of (2) is less than 40nThen zero padding is carried out; and if the value is more than 40, the operation is cut off.
4) Calculating and extracting characteristics: frame group F using the described feature extraction proceduregAnd 4 preset types of sub-features are extracted so as to combine to obtain the 635-dimensional steganalysis feature.
5) And (5) repeatedly executing the steps 2) to 4), and sequentially carrying out steganalysis feature extraction on all frame groups of the video to be detected.
6) Steganalysis: and performing steganography classification judgment on each frame group in the video to be detected by adopting a classifier based on the 635-dimensional steganography analysis characteristics.
As can be seen from the above detailed description: firstly, the invention effectively reflects the influence of video steganography in a DCT coefficient domain on the correlation of the entropy coding code words and the DCT coefficients of the adjacent blocks by constructing the correlation of the characteristic description entropy coding code words and the nonzero DCT coefficient numbers of the adjacent blocks, and the realization of the method is not dependent on a specific video coding standard. The invention has wider application range.
In order to highlight the description, the invention provides a video steganalysis method for a high-performance DCT coefficient domain, which adopts the following experimental configuration to carry out steganalysis experiments:
1) YUV sequence: 100 video sequences of 1920 x 1080 resolution YUV420P were collected over the internet, averaging 932 frames per video sequence.
2) Video encoder and its configuration: and preparing compressed video samples by adopting an x264 open source video encoder, and setting the encoding grade as a basic grade Baseline Profile in order to reduce time overhead.
3) Compressing video parameters: a Constant Rate Factor (CRF) is set to 18, 23 or 28, and the frame Rate is set to 30 fps.
4) Steganographic embedding rate: the steganographic embedding Rate is expressed by a Modification Rate (MR, the ratio of modified DCT coefficients to all DCT coefficients as a carrier), and the MR of the carrier video is set to 0, and the MR of the steganographic video is set to 0.01, 0.02, or 0.05.
5) The steganography method comprises the following steps: the high concealment DCT coefficient domain video steganography method proposed by Cao et al was chosen for analysis (reference: Y. Cao, Y. Wang, X ZHao, M Zhu, and Z Xu, "Cover Block recoupling for Content-Adaptive H.264Steganographic," in Proc.6th ACM works shop Inf.Hihg Multimedia Security, Austria, June.2018, pp.23-30).
6) The steganalysis method comprises the following steps: VDCTR is the most efficient steganalysis method for the DCT coefficient domain at present, so it is compared with the method of the present invention (ref: P.Wang, Y.Cao, X.ZHao, and M.Zhu, "A Steganalytic Algorithm to Detect DCT-based Data Hiding Methods for H.264/AVC video," in Proc.5th ACM works shop Inf.Hiding Multimedia Security, USA, June.2017, pp.123-133).
7) Training and detecting: in each group of steganalysis experiments, 50% of carrier-steganalysis sample pairs are randomly selected for training a Support Vector Machine (SVM), the remaining 50% of sample pairs are used for steganalysis classification judgment, each group of steganalysis experiments are repeated for 20 times, and the obtained data are averaged.
According to the experimental configuration, the obtained steganography analysis results are shown in table 1, and it can be seen that the method can effectively detect the video steganography of the DCT coefficient domain with the highest steganography security at the latest development stage. In addition, the steganalysis effect of the method is obviously superior to that of a VDCTR (vertical data transformer), so that the method is very suitable for steganalysis scenes with high requirements on security level.
TABLE 1 average detection accuracy (%) -of steganalysis using VDCTR and the method of the invention proposed by Cao et al
Based on the same inventive concept, another embodiment of the present invention provides a video steganalysis device based on analyzing CAVLC codewords and non-zero DCT coefficients by using the method of the present invention, which comprises:
the frame group dividing module is used for dividing the compressed video to be detected into a plurality of frame groups;
the characteristic extraction module is used for analyzing the correlation between CAVLC entropy coding code words and the number of nonzero DCT coefficients of adjacent 4 multiplied by 4 brightness macro blocks to obtain video steganalysis characteristics for each frame group;
and the steganography analysis module is used for performing steganography classification judgment on each frame group in the compressed video to be detected by using the obtained video steganography analysis characteristics and adopting a classifier.
The specific implementation process of each module is referred to the description of the method of the invention.
Based on the same inventive concept, another embodiment of the present invention provides an electronic device (computer, server, smartphone, etc.) comprising a memory storing a computer program configured to be executed by the processor, and a processor, the computer program comprising instructions for performing the steps of the inventive method.
Based on the same inventive concept, another embodiment of the present invention provides a computer-readable storage medium (e.g., ROM/RAM, magnetic disk, optical disk) storing a computer program, which when executed by a computer, performs the steps of the inventive method.
The above embodiments are only intended to illustrate the technical solution of the present invention and not to limit the same, and a person skilled in the art can modify the technical solution of the present invention or substitute the same without departing from the spirit and scope of the present invention, and the scope of the present invention should be determined by the claims.
Claims (10)
1. A video steganalysis method based on analysis of CAVLC code words and non-zero DCT coefficient numbers is characterized by comprising the following steps:
dividing a compressed video to be detected into a plurality of frame groups;
for each frame group, obtaining video steganalysis characteristics by analyzing the correlation between CAVLC entropy coding code words and the number of nonzero DCT coefficients of adjacent 4 multiplied by 4 brightness macro blocks;
and performing steganography classification judgment on each frame group in the compressed video to be detected by using the obtained video steganography analysis characteristics and adopting a classifier.
2. The method of claim 1, wherein the dividing the compressed video to be tested into a number of frame groups comprises: each frame group consists of continuous video frames, and the video frames of each frame group are not overlapped with each other and have the same number.
3. The method of claim 1, wherein obtaining the video steganalysis characteristics by analyzing the correlation between CAVLC entropy coded codewords and the number of nonzero DCT coefficients of adjacent 4 x 4 luminance macroblocks for each frame group comprises:
1) for frame group FgEach 4 x 4 luminance macroblock in the block obtains its corresponding CAVLC codeword Bn，n∈[1,N]；
2) Decoding CAVLC codeword BnRecording the relative position (x, y) of the macro block in the video frame and the corresponding number of the nonzero DCT coefficients, and calculating the correlation of the number of the nonzero DCT coefficients of the adjacent macro blocks; wherein
H is the width of the video frame, and W is the length of the video frame;
3) traversal frame set FgAll CAVLC code words B innCalculating the proportion of '1' in the code words under different numbers of the nonzero DCT coefficients and the proportion of '1' in each position of all the code words;
4) according to the calculation results of the step 2) and the step 3), the frame group F is setgAnd extracting video steganalysis characteristics.
4. The method as claimed in claim 3, wherein in step 1), for the video adopting non-CAVLC entropy coding standard, decoding is performed to obtain its corresponding DCT coefficient, and then CAVLC coding is performed to obtain corresponding CAVLC codeword Bn。
5. The method of claim 3Characterised in that step 4) consists in applying to a certain group of frames FgExtracting four types of sub-features and finally forming to obtain 635-dimensional steganalysis features; the four types of sub-features include:
a type 1 sub-feature, each dimension of which corresponds to the proportion of '1' in the codeword given the number k of non-zero DCT coefficients;
a type 2 sub-feature, each dimension of which corresponds to the proportion of "1" at the kth position in the codeword;
a type 3 sub-feature relating to the correlation of the number of DCT coefficients of upper and lower adjacent blocks, each dimension corresponding to the probability that the number of DCT coefficients of an adjacent lower block equals j given the number of DCT coefficients of an upper block equals i;
and the type 4 sub-feature is related to the correlation of the number of the nonzero DCT coefficients of the left and right adjacent blocks, and the probability that the number of the nonzero DCT coefficients of the adjacent right block is equal to j under the condition that the number of the nonzero DCT coefficients of the given left block corresponding to each dimension is equal to i.
6. Method according to claim 5, characterized in that step 4) is performed according to the following steps for a certain group of frames FgExtracting four types of sub-features and finally composing to obtain 635-dimensional steganalysis features:
a) for FgEach 4 × 4 luma macroblock in (B) gets its corresponding CAVLC codeword BnDecoding CAVLC code word to obtain the number of non-zero DCT coefficients
Wherein N is an element of [1, N ∈],
(x, y) is the relative position of the block in the video frame,
h is the width of the video frame, and W is the length of the video frame; statistics BnNumber of middle "1
And the number of "0
And record BnCode word at each position in
Wherein the content of the first and second substances,
i∈[1,40]if B isnIs automatically in B when the length of (2) is less than 40nThen zero padding is carried out; if the ratio is more than 40, cutting off;
b) extracting a type 1 sub-feature, the type 1 sub-feature being defined as:
c) extracting type 2 sub-features, the type 2 sub-features being defined as:
wherein the function
The same applies below;
d) extracting type 3 sub-features, the type 3 sub-features being defined as:
e) extracting type 4 sub-features, the type 4 sub-features being defined as:
f) the final 635-dimensional steganalysis feature is obtained by combining the above 4 types of sub-features.
7. The method of claim 6, wherein the 635-dimensional steganalysis features are defined as:
where F (k) represents a 635-dimensional steganalysis feature.
8. A video steganalysis device based on analysis of CAVLC code words and number of non-zero DCT coefficients using the method of any of claims 1 to 7, comprising:
the frame group dividing module is used for dividing the compressed video to be detected into a plurality of frame groups;
the characteristic extraction module is used for analyzing the correlation between CAVLC entropy coding code words and the number of nonzero DCT coefficients of adjacent 4 multiplied by 4 brightness macro blocks to obtain video steganalysis characteristics for each frame group;
and the steganography analysis module is used for performing steganography classification judgment on each frame group in the compressed video to be detected by using the obtained video steganography analysis characteristics and adopting a classifier.
9. An electronic apparatus, comprising a memory and a processor, the memory storing a computer program configured to be executed by the processor, the computer program comprising instructions for performing the method of any of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which, when executed by a computer, implements the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011253138.4A CN112637605B (en) | 2020-11-11 | 2020-11-11 | Video steganalysis method and device based on analysis of CAVLC code words and number of nonzero DCT coefficients |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011253138.4A CN112637605B (en) | 2020-11-11 | 2020-11-11 | Video steganalysis method and device based on analysis of CAVLC code words and number of nonzero DCT coefficients |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112637605A true CN112637605A (en) | 2021-04-09 |
| CN112637605B CN112637605B (en) | 2022-01-11 |
Family
ID=75304339
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011253138.4A Active CN112637605B (en) | 2020-11-11 | 2020-11-11 | Video steganalysis method and device based on analysis of CAVLC code words and number of nonzero DCT coefficients |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112637605B (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010068421A (en) * | 2008-09-12 | 2010-03-25 | Mitsubishi Electric Corp | Digital watermark apparatus and digital watermark method |
| CN101841700A (en) * | 2010-05-18 | 2010-09-22 | 宁波大学 | H.264 compression bit stream oriented video watermark method |
| CN103152578A (en) * | 2013-03-25 | 2013-06-12 | 中国传媒大学 | H.264 video watermark embedding and extraction method based on mixed coding/decoding |
| CN104301733A (en) * | 2014-09-06 | 2015-01-21 | 南京邮电大学 | Video steganalysis method based on feature fusions |
| CN106203492A (en) * | 2016-06-30 | 2016-12-07 | 中国科学院计算技术研究所 | The system and method that a kind of image latent writing is analyzed |
| US20170034523A1 (en) * | 2015-07-31 | 2017-02-02 | Ecole Polytechnique Federale De Lausanne (Epfl) | Media content processing method |
| CN106713917A (en) * | 2016-12-05 | 2017-05-24 | 南京航空航天大学 | Video steganography algorithm based on motion vector difference |
| CN107197297A (en) * | 2017-06-14 | 2017-09-22 | 中国科学院信息工程研究所 | A kind of video steganalysis method of the detection based on DCT coefficient steganography |
| CN108965887A (en) * | 2018-06-29 | 2018-12-07 | 中国科学院信息工程研究所 | A kind of video information hiding method and device based on uncoupling between block |
| CN111263157A (en) * | 2020-02-27 | 2020-06-09 | 武汉大学 | Video multi-domain steganalysis method based on motion vector consistency |
| CN111696021A (en) * | 2020-06-10 | 2020-09-22 | 中国人民武装警察部队工程大学 | Image self-adaptive steganalysis system and method based on significance detection |
-
2020
- 2020-11-11 CN CN202011253138.4A patent/CN112637605B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010068421A (en) * | 2008-09-12 | 2010-03-25 | Mitsubishi Electric Corp | Digital watermark apparatus and digital watermark method |
| CN101841700A (en) * | 2010-05-18 | 2010-09-22 | 宁波大学 | H.264 compression bit stream oriented video watermark method |
| CN103152578A (en) * | 2013-03-25 | 2013-06-12 | 中国传媒大学 | H.264 video watermark embedding and extraction method based on mixed coding/decoding |
| CN104301733A (en) * | 2014-09-06 | 2015-01-21 | 南京邮电大学 | Video steganalysis method based on feature fusions |
| US20170034523A1 (en) * | 2015-07-31 | 2017-02-02 | Ecole Polytechnique Federale De Lausanne (Epfl) | Media content processing method |
| CN106203492A (en) * | 2016-06-30 | 2016-12-07 | 中国科学院计算技术研究所 | The system and method that a kind of image latent writing is analyzed |
| CN106713917A (en) * | 2016-12-05 | 2017-05-24 | 南京航空航天大学 | Video steganography algorithm based on motion vector difference |
| CN107197297A (en) * | 2017-06-14 | 2017-09-22 | 中国科学院信息工程研究所 | A kind of video steganalysis method of the detection based on DCT coefficient steganography |
| CN108965887A (en) * | 2018-06-29 | 2018-12-07 | 中国科学院信息工程研究所 | A kind of video information hiding method and device based on uncoupling between block |
| CN111263157A (en) * | 2020-02-27 | 2020-06-09 | 武汉大学 | Video multi-domain steganalysis method based on motion vector consistency |
| CN111696021A (en) * | 2020-06-10 | 2020-09-22 | 中国人民武装警察部队工程大学 | Image self-adaptive steganalysis system and method based on significance detection |
Non-Patent Citations (1)
| Title |
|---|
| 张弘等: "视频隐写分析技术研究综述", 《信息安全学报》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112637605B (en) | 2022-01-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108632625B (en) | Video coding method, video decoding method and related equipment | |
| US7936824B2 (en) | Method for coding and decoding moving picture | |
| US8064516B2 (en) | Text recognition during video compression | |
| US20110299604A1 (en) | Method and apparatus for adaptive video sharpening | |
| EP1359770B1 (en) | Signaling for fading compensation in video encoding | |
| US20110222597A1 (en) | Method and apparatus for sparsity-based de-artifact filtering for video encoding and decoding | |
| US8780971B1 (en) | System and method of encoding using selectable loop filters | |
| CN108347612B (en) | Monitoring video compression and reconstruction method based on visual attention mechanism | |
| EP2614645A1 (en) | Video decoding using example - based data pruning | |
| CN111837389A (en) | Block detection method and device suitable for multi-sign bit hiding | |
| CN112400323B (en) | Image encoder, image decoder, and corresponding methods | |
| EP1845729A1 (en) | Transmission of post-filter hints | |
| WO2017004889A1 (en) | Jnd factor-based super-pixel gaussian filter pre-processing method | |
| US20150249829A1 (en) | Method, Apparatus and Computer Program Product for Video Compression | |
| CN101540900A (en) | Method for reducing block effect in video streaming | |
| Ma et al. | Rate-perception optimized preprocessing for video coding | |
| CN112637605B (en) | Video steganalysis method and device based on analysis of CAVLC code words and number of nonzero DCT coefficients | |
| CN115002482B (en) | End-to-end video compression method and system using structural preserving motion estimation | |
| CN112272297B (en) | Image quality still frame detection method embedded in decoder | |
| CN104410863B (en) | Image processor and image processing method | |
| US20120133836A1 (en) | Frame level quantization estimation | |
| CN117136540A (en) | Residual coding method and device, video coding method and device, and storage medium | |
| KR101247024B1 (en) | Method of motion estimation and compensation using in-loop preprocessing filtering | |
| CN111212288A (en) | Video data encoding and decoding method and device, computer equipment and storage medium | |
| Tang et al. | Optimization of CU Partition Based on Texture Degree in H. 266/VVC |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | 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 |