Video steganography method and system for resisting motion vector local optimum and consistency characteristics

Abstract

The invention discloses a video steganography method and system for resisting motion vector local optimum and consistency characteristics, and the method comprises the steps: obtaining a carrier video, and carrying out the segmentation of the carrier video, and obtaining a plurality of video segments with a set length; segmenting secret information to be embedded according to the capacity of the video clip to obtain a plurality of secret information bit clips; constructing a steganography distortion cost function based on the local optimality of the motion vector, the consistency of the motion vector, the eigenenergy of the block and the objective coding performance of the video; in combination with a steganography distortion cost function, embedding the secret information bit segments into the corresponding video segments to obtain a plurality of secret-carrying video segments; and splicing all the secret-carrying video clips to obtain a final secret-carrying video. According to the method, when the steganography distortion cost function is constructed, various influence factors are comprehensively considered, so that the detection of various video steganography analysis methods can be resisted while the video coding performance is ensured, and the video steganography safety and concealment are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of video steganography, and in particular to a video steganography method and system that resists local optimum and consistency features of motion vectors. Background Art

[0002] At present, with the rapid development of network and information technology, multimedia resources are widely disseminated and shared on the Internet, and the information security problems caused by this are becoming increasingly serious. Multimedia information steganography technology is a technology specifically designed for such information security problems. Multimedia information steganography technology is to embed secret information into multimedia in a certain way without causing suspicion from monitoring third parties. At present, the mainstream multimedia information steganography technology generally uses images as carriers. However, due to the limited hiding capacity of images and the easy suspicion of third parties, image steganography methods can no longer meet normal needs. In this context, video steganography technology, as an important branch of multimedia information steganography technology, has gradually received widespread attention and research. Compared with image steganography, video steganography has higher hiding capacity and stronger concealment. Video consists of continuous image frames, each of which contains a large amount of information, providing a broad space for the embedding of secret information. At the same time, the natural characteristics and dynamic changes of videos also provide better cover for the hiding of steganographic information, making it more difficult for monitoring third parties to detect the existence of steganography. In H.264 / AVC video steganography technology, modifying the motion vector value to achieve information embedding has become the mainstream method. There are two main advantages of using motion vectors as carriers: first, the richness of motion vectors in digital videos makes large-capacity video steganography possible; second, the impact of modifying motion vectors on video encoding is relatively small, which helps to achieve high-security steganography. For this reason, H.264 / AVC video steganography technology based on motion vectors has achieved remarkable research results. The video steganography algorithm based on motion vectors can be summarized as follows:

[0003] In early studies, heuristic algorithms were mostly used for video steganography based on motion vectors. Such algorithms usually set a threshold, and only when the target motion vector exceeds this threshold, information is embedded. Xu et al. embedded secret information into motion vectors that exceed the threshold through LSB matching technology. However, this method failed to fully consider the motion complexity of the video in threshold setting, resulting in unsatisfactory visual quality of the secret video. To improve this, Fang et al. proposed embedding information in motion vectors whose phase exceeds the threshold from the perspective of motion vector phase value to better reflect the motion characteristics of the video. Aly et al. further suggested embedding information in motion vectors with large prediction residuals. The algorithm proposed by Aly et al. is the most secure algorithm among such heuristic algorithms. However, these heuristic methods focus too much on the visual quality of the video and ignore the ability to resist steganalysis. With the introduction of the minimum distortion embedding framework, adaptive steganography schemes came into being, bringing new breakthroughs to H.264 / AVC video steganography technology based on motion vectors. Yao et al. conducted in-depth and systematic research, discovered the relationship between the statistical distribution of motion vectors and the change of prediction residuals, and designed the distortion cost accordingly. These steganographic algorithms reduce the modification rate of motion vectors by combining matrix coding, thus achieving good security performance in resisting detection by steganalysis methods based on motion vector correlation. The disadvantage of the method proposed by Yao et al. is that it only considers the correlation between motion vectors, but does not consider the local optimality and consistency characteristics of motion vectors. Therefore, it has poor performance against steganalysis methods NPELO and MVC based on local optimality of motion vectors. In order to resist the local optimality characteristics of motion vectors, Zhang et al. proposed a new video steganography algorithm. They searched for the local optimal motion vector within a set range, and selected the motion vector with the smallest absolute difference between the Lagrangian cost and the carrier motion vector as the optimal choice, and used this absolute difference as the distortion cost. This method specifically designs the distortion cost according to the steganalysis features with the best detection performance, thereby achieving higher security performance. However, this type of algorithm has poor security performance when facing the detection of motion vector consistency features. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a video steganography method that can improve the security and concealment of video steganography and resist the local optimum and consistency characteristics of motion vectors.

[0005] To achieve the above purpose, the technical solution provided by the present invention is:

[0006] The video steganography method against the local optimum and consistency features of motion vectors includes:

[0007] Obtain a carrier video and segment the carrier video to obtain a number of video clips of set lengths;

[0008] The secret information to be embedded is divided according to the capacity of the video segment to obtain a number of secret information bit segments;

[0009] Based on the local optimality of motion vectors, the consistency of motion vectors, the intrinsic energy of blocks and the objective coding performance of videos, a steganalysis distortion cost function is constructed.

[0010] Combined with the steganalysis distortion cost function, the secret information bit fragments are embedded into the corresponding video fragments to obtain multiple secret video fragments;

[0011] All the secret-carrying video clips are spliced ​​together to get the final secret-carrying video.

[0012] In this technical solution, when constructing the steganalysis distortion cost function, a variety of influencing factors are comprehensively considered, so as to achieve the goal of resisting the detection of various video steganalysis methods while ensuring the video encoding performance, thereby improving the security and concealment of video steganography.

[0013] In addition, the segmented embedding strategy can embed secret messages in both the horizontal and vertical components of the motion vector, thereby improving video coding performance and steganographic imperceptibility, making the secret video more difficult to detect and identify.

[0014] Furthermore, the process of constructing the steganalysis distortion cost function includes:

[0015] Based on the local optimality of motion vectors and the objective coding performance of the video, the motion vector local optimality distortion cost function ρ is constructed i (LO) ;

[0016] Based on the consistency of motion vectors and the objective coding performance of the video, a motion vector consistency distortion cost function ρ is constructed i (dMVC) ;

[0017] The motion vector local optimality distortion cost function ρ i (LO) And the motion vector consistency distortion cost function ρ i (dMVC) Combined with, the steganalysis distortion cost function ρ is constructed i .

[0018] Furthermore, the motion vector local optimality distortion cost function ρ is constructed i (LO) The process includes:

[0019] Get one of the video clips and extract the motion vector of the video clip, denoted as mv i =(mvx i ,mvy i )=[mv1,mv2…,mv n ], where n is the number of motion vectors in the video clip;

[0020] In its neighborhood Search for alternative motion vectors within, alternative motion vector set It is expressed by the following formula:

[0021]

[0022] Among them, LO and non-LO represent local optimality and non-local optimality respectively, mv′ i is an alternative motion vector, For mv i The set of neighborhood motion vectors;

[0023] mvx i The set of alternative motion vectors is:

[0024] Where Δx i =τ,τ∈{-1,0,1};

[0025] The proposed mvx i Directional local optimality distortion cost function ρ i (LO) Defined as:

[0026]

[0027] in, is the motion vector mv i The Lagrangian cost of is the Lagrangian cost of the neighborhood replaceable motion vector, given mvx i Modification of direction Δx i After that, the corresponding mvy i The local optimal distortion cost function is expressed as:

[0028]

[0029] in, Used to measure the objective encoding performance of a video.

[0030] Furthermore, the process of constructing the motion vector consistency degree distortion cost function includes:

[0031] Calculate the consistency degree of motion vector dMVC(·):

[0032]

[0033] Where δ(·) is the Kronecker function, which outputs 1 when the two inputs are equal and 0 otherwise; mvz i (z=x or y) represents mv i The horizontal or vertical component of; K is the number of motion vectors in the macroblock or sub-macroblock;

[0034] The modification amount is Δz i The motion vector component mvz of (z=x or y) i The consistency distortion cost function is expressed as follows:

[0035]

[0036] The intrinsic energy E of the block corresponding to the motion vector C is defined as follows:

[0037]

[0038] Among them, f i is the position of the i-th coefficient in the block after DCT transformation, w i =u i ·v i is the weight to measure the embeddability of the coefficient, (u i ,v i ) is the position of the block corresponding to the current motion vector in this frame;

[0039] Obtain the block intrinsic energy distortion measure ρ i,energy (Δz):

[0040]

[0041] Where ε = 10 -2 A stable constant to prevent division by zero;

[0042] Combined modification amount is Δz i The motion vector component mvz of (z=x or y) i The consistency distortion cost function and the block intrinsic energy distortion metric ρ i,energy (Δz), the modification amount is Δz i The motion vector component mvz of (z=x or y) i Distortion cost function of consistency degree:

[0043]

[0044] Furthermore, we construct the steganalysis distortion cost function ρ i The formula is as follows:

[0045] ρ i =(ρ i (LO) ) α ·(ρ i (dMVC) ) β

[0046] Among them, ρ i (LO) is the local optimality distortion cost function of the motion vector, ρ i (dMVC) is the motion vector consistency distortion cost function, α and β are weight factors for balancing the local optimal motion vector distortion cost function and the consistency distortion cost function.

[0047] Furthermore, the steganalysis distortion cost of each video clip is calculated through the steganalysis distortion cost function; then the steganalysis distortion cost of each video clip is applied to the minimum distortion steganography framework based on STC to guide the embedding position of the information in the video clip, so as to embed the secret information bit fragment into the corresponding video clip and obtain multiple secret video clips.

[0048] Furthermore, the present invention also provides a video steganography system that resists local optimum and consistency characteristics of motion vectors, which is used to implement the above-mentioned video steganography method that resists local optimum and consistency characteristics of motion vectors, and includes a carrier video acquisition module, a carrier video segmentation module, a secret information segmentation module, a steganalysis distortion cost function construction module, a secret information bit segment embedding module, and a video splicing module;

[0049] Wherein, the carrier video acquisition module is used to acquire the carrier video;

[0050] The carrier video segmentation module is used to segment the carrier video to obtain a number of video segments of set lengths;

[0051] The secret information segmentation module is used to segment the secret information to be embedded according to the capacity of the video segment to obtain a number of secret information bit segments;

[0052] The steganalysis distortion cost function construction module is used to construct a steganalysis distortion cost function based on the local optimality of the motion vector, the consistency of the motion vector, the intrinsic energy of the block and the objective coding performance of the video;

[0053] The secret information bit segment embedding module is used to embed the secret information bit segment into the corresponding video segment in combination with the steganographic distortion cost function to obtain multiple secret-carrying video segments;

[0054] The video splicing module is used to splice all the classified video clips to obtain the final classified video.

[0055] Furthermore, the present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned video steganography method for resisting local optimality and consistency characteristics of motion vectors when executing the computer program.

[0056] Furthermore, the present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the video steganography method for resisting the local optimum and consistency characteristics of motion vectors are implemented.

[0057] Compared with the prior art, the principles and advantages of this technical solution are as follows:

[0058] 1. When constructing the steganalysis distortion cost function, a variety of influencing factors are comprehensively considered, so as to ensure that the video encoding performance can be resisted by a variety of video steganalysis methods, thereby improving the security and concealment of video steganography.

[0059] 2. When constructing the motion vector local optimality distortion cost function and the motion vector consistency degree distortion cost function, the probability of the modified motion vector maintaining local optimality and consistency can be increased through local optimality judgment and consistency degree judgment, thereby further enhancing the security and concealment of steganography and reducing the risk of being detected by steganalysis algorithms.

[0060] 3. The segmented embedding strategy can embed secret messages in both the horizontal and vertical components of the motion vector, thereby improving video coding performance and steganographic imperceptibility, making the secret video more difficult to detect and identify. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the services required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0062] Figure 1 It is a structural block diagram of the video steganography system of the present invention that resists the local optimum and consistency characteristics of motion vectors;

[0063] Figure 2 The figure is a flow chart showing the principle of the video steganography method for resisting the local optimum and consistency characteristics of the motion vector according to the present invention. DETAILED DESCRIPTION

[0064] The present invention will be further described below in conjunction with specific embodiments:

[0065] like Figure 1 As shown, the video steganography system against local optimum and consistency characteristics of motion vectors described in this embodiment includes a carrier video acquisition module, a carrier video segmentation module, a secret information segmentation module, a steganography distortion cost function construction module, a secret information bit segment embedding module, and a video splicing module;

[0066] like Figure 2 As shown in FIG. 1 , the working principle of the video steganography system against the local optimum and consistency features of motion vectors includes the following steps:

[0067] S1, obtaining a carrier video through a carrier video acquisition module, and segmenting the carrier video through a carrier video segmentation module to obtain a number of video clips of set lengths;

[0068] S2, dividing the secret information to be embedded according to the capacity of the video segment by a secret information dividing module to obtain a plurality of secret information bit segments;

[0069] S3, constructing a steganalysis distortion cost function based on the local optimality of the motion vector, the consistency of the motion vector, the intrinsic energy of the block, and the objective coding performance of the video through the steganalysis distortion cost function construction module;

[0070] In this step, the process of constructing the steganalysis distortion cost function includes:

[0071] S3-1. Based on the local optimality of motion vectors and the objective coding performance of the video, a motion vector local optimality distortion cost function ρ is constructed i (LO) , the specific steps are as follows:

[0072] Get one of the video clips and extract the motion vector of the video clip, denoted as mv i =(mvx i ,mvy i )=[mv1,mv2…,mv n ], where n is the number of motion vectors in the video clip;

[0073] In its neighborhood Search for alternative motion vectors within, alternative motion vector set It is expressed by the following formula:

[0074]

[0075] Among them, LO and non-LO represent local optimality and non-local optimality respectively, mv′ iis an alternative motion vector, For mv i The set of neighborhood motion vectors;

[0076] mvx i The set of alternative motion vectors is:

[0077] Where Δx i =τ,τ∈{-1,0,1};

[0078] The proposed mvx i Directional local optimality distortion cost function ρ i (LO) Defined as:

[0079]

[0080] in, is the motion vector mv i The Lagrangian cost of is the Lagrangian cost of the neighborhood replaceable motion vector, given mvx i Modification of direction Δx i After that, the corresponding mvy i The local optimal distortion cost function is expressed as:

[0081]

[0082] in, Used to measure the objective encoding performance of a video.

[0083] S3-2. Based on the consistency of motion vectors and the objective coding performance of the video, a motion vector consistency distortion cost function ρ is constructed. i (dMVC) , the specific steps are as follows:

[0084] Calculate the consistency degree of motion vector dMVC(·):

[0085]

[0086] Where δ(·) is the Kronecker function, which outputs 1 when the two inputs are equal and 0 otherwise; mvz i (z=x or y) represents mv i The horizontal or vertical component of; K is the number of motion vectors in the macroblock or sub-macroblock;

[0087] The modification amount is Δz i The motion vector component mvz of (z=x or y) i The consistency distortion cost function is expressed as follows:

[0088]

[0089] The intrinsic energy E of the block corresponding to the motion vector C is defined as follows:

[0090]

[0091] Among them, f i is the position of the i-th coefficient in the block after DCT transformation, w i =u i ·v i is the weight to measure the embeddability of the coefficient, (u i ,v i ) is the position of the block corresponding to the current motion vector in this frame;

[0092] Obtain the block intrinsic energy distortion measure ρ i,energy (Δz):

[0093]

[0094] Where ε = 10 -2 A stable constant to prevent division by zero;

[0095] Combined modification amount is Δz i The motion vector component mvz of (z=x or y) i The consistency distortion cost function and the block intrinsic energy distortion metric ρ i,energy (Δz), the modification amount is Δz i The motion vector component mvz of (z=x or y) i Distortion cost function of consistency degree:

[0096]

[0097] S3-3, the motion vector local optimality distortion cost function ρ i (LO) And the motion vector consistency distortion cost function ρ i (dMVC) Combined with, the steganalysis distortion cost function ρ is constructed i .

[0098] Constructing the steganalysis distortion cost function ρ i The formula is as follows:

[0099] ρ i =(ρ i (LO) ) α ·(ρ i(dMVC) ) β

[0100] Among them, ρ i (LO) is the local optimality distortion cost function of the motion vector, ρ i (dMVC) is the motion vector consistency distortion cost function, α and β are weight factors for balancing the local optimal motion vector distortion cost function and the consistency distortion cost function.

[0101] S4. The steganalysis distortion cost of each video segment is calculated by the steganalysis distortion cost function; the steganalysis distortion cost of each video segment is then applied to the minimum distortion steganography framework based on STC to guide the embedding position of the information in the video segment, thereby embedding the secret information bit segment into the corresponding video segment to obtain multiple secret video segments.

[0102] S5. All the secret-carrying video clips are spliced ​​together through the video splicing module to obtain the final secret-carrying video.

[0103] The best current motion vector-based steganography algorithm can only resist a single steganalysis method. To address this technical problem, this embodiment comprehensively considers aspects such as the local optimality of the motion vector, the consistency of the motion vector, the intrinsic energy of the block, and the objective coding performance of the video, and reasonably integrates these influencing factors into a cost function, so that the method described in this embodiment can simultaneously resist multiple video steganalysis algorithms while ensuring the objective coding performance of the video.

[0104] The existing method proposed by Zhang et al. uses prediction error to characterize the local optimality of motion vectors and constructs its steganalysis distortion cost function. The steganalysis security performance under this method usually depends to a large extent on the prediction accuracy of motion estimation. However, this embodiment abandons the idea of ​​using prediction error to characterize the local optimality of motion vectors, and instead uses Lagrangian rate distortion cost to characterize the local optimality of motion vectors, and creates a new steganalysis distortion cost function. This saves the algorithm the trouble of over-reliance on the prediction accuracy of motion estimation, thereby achieving higher steganalysis security.

[0105] In order to verify that the steganographic method of this embodiment has higher steganographic security, a steganographic test was carried out on a test library consisting of 30 standard video sequences based on the above technical solution. The 30 video sequences were divided into 100 non-overlapping subsequences of 100 frames in length. During the test, all steganographic schemes were simulated under their corresponding load-distortion limits, with a relative load γ∈[0.1-0.3]bpnsmv. For the design of the steganalyzer, the currently optimal steganographic features NPELO and MVC are used to evaluate the empirical security performance of the steganographic scheme. 50% of the carrier / secret video pairs in the data set are used to train the classifier, and the other half will be used as a test set. The security performance is the average classification error probability of 20 random tests, denoted as

[0106] In order to reflect the advantages of the proposed scheme, three mainstream video steganography methods are used for comparative experiments, namely, the video steganography method based on motion vector prediction residual proposed by Aly et al. (denoted as Tar1), the video steganography method based on local optimality of motion vector proposed by Zhang et al. (denoted as Tar2), and the video steganography method based on the change of statistical distribution of motion vector and prediction error proposed by Yao et al. (denoted as Tar3). The experimental results are shown in Tables 1 and 2 below. It can be found from the table that the performance of Tar1, Tar2 and Tar3 methods against different steganalysis features is not stable and fluctuates greatly. The scheme proposed in this embodiment has achieved the best performance against almost all tested steganalysis features.

[0107] Table 1 Average classification error probability of the steganographic method LO_dMVC of the present invention and the steganographic methods Tar1-3 of the prior art against MVC features at different embedding rates

[0108]

[0109] Table 2 Average classification error probability of the steganographic method LO_dMVC of the present invention and the steganographic methods Tar1-3 of the prior art against NPELO features at different embedding rates

[0110]

[0111] In addition to steganographic security, video coding performance is also an important indicator for evaluating steganographic quality. In order to compare the video coding performance of the steganographic schemes involved, 100 video clips were selected from the video database, and then the coding performance of the tested steganographic schemes under different loads was compared. As shown in Table 3 below, the comparison results show that the video coding performance of the LO_dMVC scheme proposed in the present invention is better, meeting normal video coding requirements.

[0112] Table 3 Video coding performance of the steganographic method LO_dMVC of the present invention and the steganographic methods Tar1 to 3 of the prior art at different embedding rates.

[0113]

[0114]

[0115] in addition,

[0116] This embodiment also includes a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the video steganography method for resisting the local optimum and consistency characteristics of motion vectors are implemented.

[0117] This embodiment also includes a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the video steganography method for resisting the local optimum and consistency characteristics of motion vectors are implemented.

[0118] The embodiments described above are only preferred embodiments of the present invention and are not intended to limit the scope of implementation of the present invention. Therefore, all changes made according to the shape and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A video steganography method against local optimum and consistency features of motion vectors, characterized in that: include: Obtain a carrier video and segment the carrier video to obtain a number of video clips of set lengths; The secret information to be embedded is divided according to the capacity of the video segment to obtain a number of secret information bit segments; Based on the local optimality of motion vectors, the consistency of motion vectors, the intrinsic energy of blocks and the objective coding performance of videos, a steganalysis distortion cost function is constructed. Combined with the steganalysis distortion cost function, the secret information bit fragments are embedded into the corresponding video fragments to obtain multiple secret video fragments; All the secret-carrying video clips are spliced ​​together to get the final secret-carrying video.

2. The video steganography method against local optimum and consistency features of motion vector according to claim 1, characterized in that: The process of constructing the steganalysis distortion cost function includes: Based on the local optimality of motion vectors and the objective coding performance of the video, the motion vector local optimality (LO) distortion cost function ρ is constructed i (LO) ; Based on the consistency of motion vectors and the objective coding performance of the video, a motion vector consistency degree distortion cost function ρ is constructed i (dMVC) ; The motion vector local optimality distortion cost function ρ i (LO) And the motion vector consistency distortion cost function ρ i (dMVC) Combined with, the steganalysis distortion cost function ρ is constructed i .

3. The video steganography method against local optimum and consistency features of motion vector according to claim 2, characterized in that: Construct motion vector local optimality distortion cost function ρ i (LO) The process includes: Get one of the video clips and extract the motion vector of the video clip, denoted as mv i =(mvx i ,mvy i )=[mv1,mv2…,mv n ], where n is the number of motion vectors in the video clip; In its neighborhood Search for alternative motion vectors within, alternative motion vector set It is expressed by the following formula: Among them, LO and non-LO represent local optimality and non-local optimality respectively, mv′ i is an alternative motion vector, For mv i The set of neighborhood motion vectors; mvx i The set of alternative motion vectors is: Among them, Δx i =τ,τ∈{-1,0,1}; The proposed mvx i Directional local optimality distortion cost function ρ i (LO) Defined as: in, is the motion vector mv i The Lagrangian cost of is the Lagrangian cost of the neighborhood replaceable motion vector, given mvx i Modification of direction Δx i After that, the corresponding mvy i The local optimal distortion cost function is expressed as: in, Used to measure the objective encoding performance of a video.

4. The video steganography method against local optimum and consistency features of motion vectors according to claim 2, characterized in that: The process of constructing the motion vector consistency distortion cost function includes: Calculate the consistency degree of motion vector dMVC(·): Where δ(·) is the Kronecker function, which outputs 1 when the two inputs are equal and 0 otherwise; mvz i (z=xory) represents mv i The horizontal or vertical component of; K is the number of motion vectors in the macroblock or sub-macroblock; The modification amount is Δz i The motion vector component mvz of (z=xory) i The consistency distortion cost function is expressed as follows: The intrinsic energy E of the block corresponding to the motion vector C is defined as follows: Among them, f i is the position of the i-th coefficient in the block after DCT transformation, w i =u i ·v i is the weight to measure the embeddability of the coefficient, (u i ,v i ) is the position of the block corresponding to the current motion vector in this frame; Obtain the block intrinsic energy distortion measure ρ i,energy (Δz): Where ε = 10 -2 A stable constant to prevent division by zero; Combined modification amount is Δz i The motion vector component mvz of (z=xory) i The consistency distortion cost function and the block intrinsic energy distortion metric ρ i,energy (Δz), the modification amount is Δz i The motion vector component mvz of (z=xory) i Distortion cost function of consistency degree:

5. The video steganography method against local optimum and consistency features of motion vectors according to claim 2, characterized in that: Constructing the steganalysis distortion cost function ρ i The formula is as follows: r i =(ρ i (LO) ) α ·(r i (dMVC) ) β Among them, ρ i (LO) is the local optimality distortion cost function of the motion vector, ρ i (dMVC) is the motion vector consistency distortion cost function, α and β are weight factors for balancing the local optimal distortion cost function and the consistency distortion cost function of the motion vector.

6. The video steganography method against local optimum and consistency features of motion vectors according to claim 1, characterized in that: The steganalysis distortion cost of each video segment is calculated by the steganalysis distortion cost function; then the steganalysis distortion cost of each video segment is applied to the minimum distortion steganography framework based on STC to guide the embedding position of the information in the video segment, so as to embed the secret information bit segment into the corresponding video segment and obtain multiple secret video segments.

7. A video steganography system with local optimal and consistent characteristics of motion vectors, characterized in that the video steganography method with local optimal and consistent characteristics of motion vectors used to implement any one of claims 1 to 6 comprises a carrier video acquisition module, a carrier video segmentation module, a secret information segmentation module, a steganalysis distortion cost function construction module, a secret information bit segment embedding module, and a video splicing module; Wherein, the carrier video acquisition module is used to acquire the carrier video; The carrier video segmentation module is used to segment the carrier video to obtain a number of video clips of set lengths; The secret information segmentation module is used to segment the secret information to be embedded according to the capacity of the video segment to obtain a number of secret information bit segments; The steganalysis distortion cost function construction module is used to construct a steganalysis distortion cost function based on the local optimality of the motion vector, the consistency of the motion vector, the intrinsic energy of the block and the objective coding performance of the video; The secret information bit segment embedding module is used to embed the secret information bit segment into the corresponding video segment in combination with the steganographic distortion cost function to obtain multiple secret-carrying video segments; The video splicing module is used to splice all the classified video clips to obtain the final classified video.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the video steganography method against local optimum and consistency features of motion vectors as claimed in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the video steganography method against local optimum and consistency features of motion vectors as claimed in any one of claims 1 to 6 are implemented.

Images