CN116233441A - H.265/HEVC video self-adaptive steganography method based on improved RDO - Google Patents

Abstract

The invention discloses an H.265/HEVC video self-adaptive steganography method based on improved RDO, wherein secret information embedding comprises three parts of extracting carriers, calculating distortion cost and STC self-adaptive embedding, and according to the block size and the characteristics of PU dividing modes, proper PU dividing modes are extracted from all encoding units with the size smaller than 64 multiplied by 64 in P frames in a classified manner to serve as carriers, the distortion cost of the carriers in the simulated embedding is calculated, and according to the distortion cost, the data hiding mode with the minimum total distortion cost is realized by using STC algorithm to realize self-adaptive steganography information embedding for the carriers; because the method comprehensively considers factors such as video bit rate increase, video picture quality, inter-frame distortion transfer and the like when calculating the distortion cost value, the bit rate increase caused by steganography can be effectively reduced, and the steganographically-contained video has better picture quality, lower code rate and higher safety.

Description

Adaptive steganography method for H.265/HEVC video based on improved RDO

Technical Field

The present invention relates to a video steganography technology, and in particular to an H.265/HEVC video adaptive steganography method based on improved RDO (Rate Distortion Optimization).

Background Art

With the rapid development of digital media technology, the popularity of multimedia processing technology has made multimedia an important means for people to obtain information, and video data has exploded, especially in social networks and the entertainment industry, where video has become more and more mainstream media. While people are enjoying the dividends brought by the information age, they are also paying more and more attention to the security of electronic information content transmitted by multimedia in public channels.

In order to solve these security problems, researchers have proposed information hiding and digital encryption. Information hiding technology includes digital watermarking and steganography. Information hiding technology has been applied to many fields, such as ownership claim protection, privacy protection and covert communication. Video steganography is closely related to video coding standards. High Efficiency Video Coding (HEVC) is a video coding standard released by ITU-TVCEG and ISO/IEC MPEG. HEVC can double the compression efficiency on the basis of H.264/AVC, that is, the bit rate of the video stream is reduced by 50% while ensuring the same video image quality. HEVC video steganography technology can naturally hide the communication caused by user behavior, ensure the rationality of user behavior, and reduce the risk of exposing hidden communications. However, most of the existing HEVC video steganography technologies based on PU (Prediction Unit) have problems such as serious bit rate increase, sacrificing bit rate for video quality, and low security of steganography algorithms. For example, Zhang et al. used the improved EMD (Exploiting Modification Direction) algorithm to improve the embedding efficiency, but it led to a serious increase in bit rate. At the same time, there is no restriction on the modification of PU partition mode. During steganography, each PU partition mode can be modified to any other PU partition mode. This arbitrary modification will also lead to a decrease in security performance. Therefore, it is of great practical significance to develop a high-security HEVC video steganography method with low bit rate increase.

Summary of the invention

The technical problem to be solved by the present invention is to provide an H.265/HEVC video adaptive steganography method based on improved RDO, which can reduce the bit rate increase caused by steganography, while improving the video visual quality and having high security.

The technical solution adopted by the present invention to solve the above technical problems is: an H.265/HEVC video adaptive steganography method based on improved RDO, including two parts: secret information embedding and secret information extraction; the characteristics are:

The specific process of embedding the secret information is as follows:

Step 1_1: Use the H.265/HEVC standard encoder to compress and encode the original video. During the compression and encoding process, the prediction mode and depth of each coding unit in each coding tree unit in each frame and the PU partition mode corresponding to each coding unit are saved. After the compression and encoding, the H.265/HEVC video is obtained; wherein the size of the coding tree unit is 64×64, and the size of the coding unit is 64×64, 32×32, 16×16, or 8×8;

Step 1_2: traverse all frames in the H.265/HEVC video in order, and define the currently traversed frame as the current frame;

Step 1_3: Determine whether the current frame is a P frame. If the current frame is a P frame, execute step 1_4; if the current frame is not a P frame, directly execute step 1_7;

Step 1_4: traverse the coding units of all sizes in the current frame in order, and define the currently traversed coding unit as the current unit;

Step 1_5: If the size of the current unit is 64×64, the current unit is not processed, and then step 1_6 is executed; if the size of the current unit is 32×32 or 16×16, the PU partition mode corresponding to the current unit is used as a carrier and classified as a first type of carrier, and the index position of the current unit in the H.265/HEVC video, the depth of the current unit, and the PU partition mode corresponding to the current unit are recorded, and then step 1_6 is executed; if the size of the current unit is 8×8, the PU partition mode corresponding to the current unit is used as a carrier and classified as a second type of carrier, and the index position of the current unit in the H.265/HEVC video, the depth of the current unit, and the PU partition mode corresponding to the current unit are recorded, and then step 1_6 is executed;

Step 1_6: traverse the next coding unit in the current frame as the current unit, then return to step 1_5 to continue executing until all coding units in the current frame are traversed, and then execute step 1_7;

Step 1_7: traverse the next frame in the H.265/HEVC video as the current frame, then return to step 1_3 to continue executing, and execute step 1_8 after all frames in the H.265/HEVC video have been traversed;

Step 1_8: Map all carriers classified as the first type of carriers and all carriers classified as the second type of carriers, that is, PU partitioning modes, respectively. The specific process is as follows:

There are eight PU partition modes for the first type of carriers, namely 2N×2N, N×N, 2N×N, 2N×nU, 2N×nD, N×2N, nL×2N, and nR×2N partition modes. The 2N×N, 2N×nU, and 2N×nD partition modes are all horizontal partition modes, and the N×2N, nL×2N, and nR×2N partition modes are all vertical partition modes. For any carrier classified as the first type of carrier, if the carrier is in the 2N×2N partition mode or the N×N partition mode, the carrier is not mapped; if the carrier is in the 2N×N partition mode, the carrier is mapped to the integer 0; if the carrier is in the 2N×nU partition mode, the carrier is mapped to the integer 0; If the carrier is a 2N×nD partition mode, the carrier is mapped to an integer 1; if the carrier is a 2N×nD partition mode, the carrier is mapped to an integer 2; if the carrier is an N×2N partition mode, the carrier is mapped to an integer 3; if the carrier is an nL×2N partition mode, the carrier is mapped to an integer 4; if the carrier is an nR×2N partition mode, the carrier is mapped to an integer 5; then all integers obtained after mapping all carriers classified as the first type of carriers are arranged in the order of the index positions of the coding units corresponding to the carrier, i.e., the PU partition mode, in the H.265/HEVC video to form a first type of carrier sequence; then execute step 1_9;

There are four PU partition modes for the second type of carrier, namely 2N×2N, 2N×N, N×2N, and N×N partition modes; for any carrier classified as the second type of carrier, if the carrier is in the 2N×2N partition mode, the carrier is not mapped; if the carrier is in the 2N×N partition mode, the carrier is mapped to an integer 0; if the carrier is in the N×2N partition mode, the carrier is mapped to an integer 1; if the carrier is in the N×N partition mode, the carrier is mapped to an integer 2; then all integers obtained after mapping all carriers classified as the second type of carrier are arranged in the order of the index positions of the coding units corresponding to the carrier, i.e., the PU partition mode, in the H.265/HEVC video to form a second type of carrier sequence; then execute step 1_9;

Step 1_9: Calculate the cost of modifying the carrier corresponding to each integer in the first type of carrier sequence to the carrier corresponding to other integers in the same type of carrier sequence, that is, the PU partition mode. For any carrier corresponding to an integer in the first type of carrier sequence, use it as the current carrier, set the current carrier to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and modify the current carrier to the carrier corresponding to any other integer t in the same type of carrier sequence, that is, the PU partition mode. The cost value is recorded as ρ ^t _m,n,k ,

Calculate the cost of changing the carrier corresponding to each integer in the second type of carrier sequence to the carrier corresponding to other integers in the same type of carrier sequence, that is, the PU partition mode. For any carrier corresponding to an integer in the second type of carrier sequence, use it as the current carrier, set the current carrier to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and change the current carrier to the carrier corresponding to any other integer s in the same type of carrier sequence, that is, the PU partition mode. The cost value is recorded as

表示将当前载体修改为同类载体序列中的其它任一个整数t对应的载体即PU划分模式后的失真，

表示将当前载体修改为同类载体序列中的其它任一个整数t对应的载体即PU划分模式后的比特数，

表示将当前载体修改为同类载体序列中的其它任一个整数t对应的载体即PU划分模式后的拉格朗日因子，

表示失真控制因子，

表示比特率控制因子，

若当前载体属于第二类载体，则当前载体对应的整数为0时s为1或2，当前载体对应的整数为1时s为0或2，当前载体对应的整数为2时s为0或1，

表示将当前载体修改为同类载体序列中的其它任一个整数s对应的载体即PU划分模式后的失真，

表示将当前载体修改为同类载体序列中的其它任一个整数s对应的载体即PU划分模式后的比特数，

表示将当前载体修改为同类载体序列中的其它任一个整数s对应的载体即PU划分模式后的拉格朗日因子；Wherein, 1≤m≤M, M represents the total number of frames contained in the H.265/HEVC video, 1≤n≤N, N represents the total number of coding tree units contained in the m-th frame in the H.265/HEVC video, 1≤n≤N, 1≤k≤K, K represents the total number of coding units contained in the n-th coding tree unit in the m-th frame in the H.265/HEVC video, if the current carrier belongs to the first type of carrier, then when the integer corresponding to the current carrier is 0, t is 1, 2, 3, 4 or 5, when the integer corresponding to the current carrier is 1, t is 0, 2, 3, 4 or 5, when the integer corresponding to the current carrier is 2, t is 0, 1, 3, 4 or 5, and the integer corresponding to the current carrier is When the integer corresponding to the current carrier is 3, t is 0, 1, 2, 4 or 5. When the integer corresponding to the current carrier is 4, t is 0, 1, 2, 3 or 5. When the integer corresponding to the current carrier is 5, t is 0, 1, 2, 3 or 4. exp() represents an exponential function with the natural base e as the base. α represents the inter-frame distortion transmission rate, 0＜α＜1. GOPSize represents the total number of frames in a picture group GOP in H.265/HEVC video. Poc represents the position of the frame where the coding unit corresponding to the current carrier is located in the picture group GOP where the frame is located. That is, the frame where the coding unit corresponding to the current carrier is located is the Poc-th frame in the picture group GOP where the frame is located. 1≤Poc≤GOPSize.

It indicates the distortion after changing the current carrier to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

Indicates the number of bits after the current carrier is modified to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

It indicates the Lagrangian factor after the current carrier is modified to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

represents the distortion control factor,

represents the bit rate control factor,

If the current carrier belongs to the second type of carrier, then when the integer corresponding to the current carrier is 0, s is 1 or 2; when the integer corresponding to the current carrier is 1, s is 0 or 2; when the integer corresponding to the current carrier is 2, s is 0 or 1.

It indicates the distortion after changing the current carrier to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode.

Indicates the number of bits after the current carrier is modified to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode.

Indicates the Lagrangian factor after the current carrier is modified to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode;

将当前整数保持不变时的失真代价值记为

将当前整数减1时的失真代价值记为

若当前整数z_m,n,k为0，则

为无穷大；若当前整数z_m,n,k为1，则

若当前整数z_m,n,k为2，则

为无穷大、

若当前整数z_m,n,k为3，则

为无穷大；若当前整数z_m,n,k为4，则

若当前整数z_m,n,k为5，则

为无穷大、

Step 1_10: Calculate the distortion cost value when each integer in the first type of carrier sequence is increased by 1, remains unchanged, or decreased by 1. For any integer in the first type of carrier sequence, take it as the current integer, set the carrier corresponding to the current integer to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and record the distortion cost value when the current integer is increased by 1 as

The distortion cost when the current integer remains unchanged is recorded as

The distortion cost when the current integer is reduced by 1 is recorded as

If the current integer z _m,n,k is 0, then

is infinite; if the current integer z _m,n,k is 1, then

If the current integer z _m,n,k is 2, then

For infinity,

If the current integer z _m,n,k is 3, then

is infinite; if the current integer z _m,n,k is 4, then

If the current integer z _m,n,k is 5, then

For infinity,

将当前整数保持不变时的失真代价值记为

将当前整数减1时的失真代价值记为

若当前整数

为0，则

为无穷大；若当前整数

为1，则

若当前整数

为2，则

为无穷大、

Calculate the distortion cost value when each integer in the second type of carrier sequence is increased by 1, remains unchanged, or decreased by 1. For any integer in the second type of carrier sequence, take it as the current integer, set the carrier corresponding to the current integer to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and record the distortion cost value when the current integer is increased by 1 as

The distortion cost when the current integer remains unchanged is recorded as

The distortion cost when the current integer is reduced by 1 is recorded as

If the current integer

If is 0,

is infinite; if the current integer

is 1, then

If the current integer

is 2, then

For infinity,

Step 1_11: Use the same embedding load rate payload to randomly generate two different binary secret information sequences, which are denoted as _X1 and _X2 respectively; then, according to the distortion cost value when each integer in the first type of carrier sequence is increased by 1, remains unchanged, or decreased by 1, use the STC toolkit to embed _X1 into all integers in the first type of carrier sequence to obtain the first type of secret carrier sequence; similarly, according to the distortion cost value when each integer in the second type of carrier sequence is increased by 1, remains unchanged, or decreased by 1, use the STC toolkit to embed _X2 into all integers in the second type of carrier sequence to obtain the second type of secret carrier sequence; where payload∈(0,0.5], the length of _X1 is _Num1 ×payload, the length of _X2 is _Num2 ×payload, _Num1 represents the total number of carriers in the first type of carrier, and _Num2 represents the total number of carriers in the second type of carrier;

Step 1_12: Convert all integers in the first type of secret carrier sequence and the second type of secret carrier sequence into PU partition patterns respectively. The specific process is as follows:

The first type of secret carrier sequence has six integers, namely 0, 1, 2, 3, 4, and 5. The integer 0 is converted into a 2N×N partitioning mode, the integer 1 is converted into a 2N×nU partitioning mode, the integer 2 is converted into a 2N×nD partitioning mode, the integer 3 is converted into an N×2N partitioning mode, the integer 4 is converted into an nL×2N partitioning mode, and the integer 5 is converted into an nR×2N partitioning mode. Then, all the PU partitioning modes obtained after converting all the integers in the first type of secret carrier sequence are used to replace the carriers in the same position in the first type of carriers in order to form the first type of secret carriers.

The second type of secret carrier sequence has three integers, namely 0, 1, 2, and 3. The integer 0 is converted into a 2N×N partition mode, the integer 1 is converted into an N×2N partition mode, and the integer 2 is converted into an N×N partition mode. Then, all the PU partition modes obtained after converting all the integers in the second type of secret carrier sequence are used to replace the carriers at the same position in the second type of carriers in order to form the second type of secret carriers.

Step 1_13: Use the H.265/HEVC standard encoder to compress and encode the original video. In the prediction process of compression encoding, the original PU partitioning pattern at the corresponding position is replaced with the first type of secret carrier and the second type of secret carrier after steganography, and the secret video code stream is obtained through compression encoding;

The specific process of extracting secret information is as follows:

Step 2_1: Use the H.265/HEVC standard decoder to decode the encrypted video stream. During the decoding process, save the prediction mode and depth of each coding unit in each coding tree unit in each frame, and the PU partition mode corresponding to each coding unit, and obtain the decoded video after decoding;

Step 2_2: Obtain the encrypted first-type carrier sequence and the encrypted second-type carrier sequence corresponding to the decoded video in the same manner as the process of obtaining the first-type carrier sequence and the second-type carrier sequence in step 1_2 to step 1_8;

同样，利用STC工具包对步骤2_2得到的含密的第二类载体序列进行解码，提取得到第二隐秘信息，记为

Step 2_3: Use the STC toolkit to decode the first type of carrier sequence obtained in step 2_2, and extract the first secret information, which is recorded as

Similarly, the STC toolkit is used to decode the secret second type carrier sequence obtained in step 2_2, and the second secret information is extracted, which is recorded as

取值为0.3，

取值为0.7。In the steps 1-9, the value of α is 0.5.

The value is 0.3.

The value is 0.7.

和

的值取决于是否采用Merge帧间预测模式，当采用Merge帧间预测模式时，

和

的值为SATD值；当采用非Merge帧间预测模式时，

和

的值为编码失真。In the steps 1-9,

and

The value of depends on whether the Merge inter-frame prediction mode is used. When the Merge inter-frame prediction mode is used,

and

The value is the SATD value; when the non-Merge inter-frame prediction mode is used,

and

The value of is the coding distortion.

和

的值取决于是否采用Merge帧间预测模式，当采用Merge帧间预测模式时，

和

的值为Merge索引的编码比特数；当采用非Merge帧间预测模式时，

和

的值为表示预测模式和残差变换系数的编码比特数。In the steps 1-9,

and

The value of depends on whether the Merge inter-frame prediction mode is used. When the Merge inter-frame prediction mode is used,

and

The value is the number of coded bits of the Merge index; when the non-Merge inter-frame prediction mode is used,

and

The value represents the number of coding bits for the prediction mode and the residual transform coefficient.

和

的值通过μ·ω·2^(QP-12)/3.0计算得到，其中，μ的取值依赖于当前载体对应的编码单元所在的帧是否为该帧所在的图像组GOP内的参考帧，若该帧为其所在的图像组GOP内的参考帧，则μ＝1.0-Clip3(0.0,0.5,0.05N_B)，若该帧为其所在的图像组GOP内的非参考帧，则μ＝1.0，N_B表示H.265/HEVC视频中一个图像组GOP内的B帧的总帧数，Clip3()为Clip操作函数，ω表示加权因子，ω的值由编码配置和当前载体对应的编码单元所在的帧在该帧所在的图像组GOP内的位置Poc决定，QP表示当前载体对应的编码单元所在的帧的编码量化参数。In the steps 1-9,

and

The value of is calculated by μ·ω·2 ^(QP-12)/3.0 , where the value of μ depends on whether the frame where the coding unit corresponding to the current carrier is located is a reference frame in the picture group GOP where the frame is located. If the frame is a reference frame in the picture group GOP where the frame is located, then μ＝1.0-Clip3( _{0.0,0.5,0.05NB} ); if the frame is a non-reference frame in the picture group GOP where the frame is located, then μ＝1.0, _NB represents the total number of B frames in a picture group GOP in the H.265/HEVC video, Clip3() is the Clip operation function, ω represents the weighting factor, and the value of ω is determined by the encoding configuration and the position Poc of the frame where the coding unit corresponding to the current carrier is located in the picture group GOP where the frame is located, and QP represents the encoding quantization parameter of the frame where the coding unit corresponding to the current carrier is located.

Compared with the prior art, the advantages of the present invention are:

1) The method of the present invention designs an adaptive distortion cost calculation function combining multiple factors (see the cost calculation formula in step 1_9) by analyzing the influence of inter-frame distortion and bit rate growth, and uses the STC algorithm to adaptively hide data in the PU partition mode (see step 1_11), which can effectively improve the video picture quality after steganography and the security of the algorithm.

2) The method of the present invention considers the decision process of the PU partition mode in HEVC encoding, refers to the influence of rate-distortion optimization (RDO) on the PU partition mode, introduces rate-distortion cost (RD cost) in the embedding process and adjusts the weight of distortion and bit rate to suppress the serious increase of bit rate caused by steganography, thereby reducing the increase of video bit rate.

3) The HEVC video steganography and extraction processes in the method of the present invention are all completed in the prediction unit. Therefore, the video code stream after data hiding is decoded using a standard HEVC video decoder. While obtaining high-definition secret video, the secret information can also be effectively extracted. This method has a wider application prospect and can be easily applied to real-time communications in real life.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a block diagram of the overall implementation of the method of the present invention;

FIG. 2 shows the encrypted video obtained by steganographically writing data on the original video using the Zhang method and the encrypted video obtained by hiding data on the original video using the method of the present invention on BRI.

DETAILED DESCRIPTION

The present invention is further described in detail below with reference to the accompanying drawings.

The present invention proposes an H.265/HEVC video adaptive steganography method based on improved RDO, and its overall implementation block diagram is shown in Figure 1, which includes two parts: secret information embedding and secret information extraction.

The specific process of embedding the secret information is as follows:

Step 1_1: Use the H.265/HEVC standard encoder to compress and encode the original video. During the compression encoding process, the prediction mode and depth of each coding unit (CU) in each coding tree unit (CTU) in each frame, and the PU (Prediction Unit) division mode corresponding to each coding unit are saved, and the H.265/HEVC video is obtained after the compression encoding is completed; wherein the size of the coding tree unit is 64×64, and the size of the coding unit is 64×64 or 32×32 or 16×16 or 8×8.

Step 1_2: Traverse all frames in the H.265/HEVC video in order, and define the currently traversed frame as the current frame.

Step 1_3: Determine whether the current frame is a P frame. If the current frame is a P frame, execute step 1_4; if the current frame is not a P frame, directly execute step 1_7.

Step 1_4: traverse coding units of all sizes in the current frame in order, and define the currently traversed coding unit as the current unit.

Step 1_5: If the size of the current unit is 64×64, the current unit is not processed, and then step 1_6 is executed; if the size of the current unit is 32×32 or 16×16, the PU partition mode corresponding to the current unit is used as a carrier and classified as a first-class carrier, and the index position of the current unit in the H.265/HEVC video, the depth of the current unit, and the PU partition mode corresponding to the current unit are recorded, and then step 1_6 is executed; if the size of the current unit is 8×8, the PU partition mode corresponding to the current unit is used as a carrier and classified as a second-class carrier, and the index position of the current unit in the H.265/HEVC video, the depth of the current unit, and the PU partition mode corresponding to the current unit are recorded, and then step 1_6 is executed.

Step 1_6: traverse the next coding unit in the current frame as the current unit, then return to step 1_5 to continue executing until all coding units in the current frame are traversed, and then execute step 1_7.

Step 1_7: Traverse the next frame in the H.265/HEVC video as the current frame, then return to step 1_3 to continue executing, and execute step 1_8 after all frames in the H.265/HEVC video have been traversed.

Step 1_8: Map all carriers classified as the first type of carriers and all carriers classified as the second type of carriers, that is, PU partitioning modes, respectively. The specific process is as follows:

There are eight PU partition modes for the first type of carriers, namely 2N×2N, N×N, 2N×N, 2N×nU, 2N×nD, N×2N, nL×2N, and nR×2N partition modes. The 2N×N, 2N×nU, and 2N×nD partition modes are all horizontal partition modes, and the N×2N, nL×2N, and nR×2N partition modes are all vertical partition modes. For any carrier classified as the first type of carrier, if the carrier is in the 2N×2N partition mode or the N×N partition mode, the carrier is not mapped; if the carrier is in the 2N×N partition mode, the carrier is mapped to the integer 0; if the carrier is in the 2N×nU partition mode, the carrier is mapped to the integer 0; If the carrier is a PU partition mode, the carrier is mapped to an integer 1; if the carrier is a 2N×nD partition mode, the carrier is mapped to an integer 2; if the carrier is an N×2N partition mode, the carrier is mapped to an integer 3; if the carrier is an nL×2N partition mode, the carrier is mapped to an integer 4; if the carrier is an nR×2N partition mode, the carrier is mapped to an integer 5; then all integers obtained after mapping all carriers classified as the first type of carriers are arranged in the order of index positions of the coding units corresponding to the carrier, i.e., the PU partition mode, in the H.265/HEVC video to form a first type of carrier sequence; then execute step 1_9.

There are four PU partition modes for the second type of carriers, namely 2N×2N, 2N×N, N×2N, and N×N partition modes; for any carrier classified as the second type of carrier, if the carrier is in the 2N×2N partition mode, the carrier is not mapped; if the carrier is in the 2N×N partition mode, the carrier is mapped to the integer 0; if the carrier is in the N×2N partition mode, the carrier is mapped to the integer 1; if the carrier is in the N×N partition mode, the carrier is mapped to the integer 2; then all integers obtained after mapping all carriers classified as the second type of carriers are arranged in the order of the index positions of the coding units corresponding to the carrier, i.e., the PU partition mode, in the H.265/HEVC video to form a second type of carrier sequence; then execute steps 1_9.

Step 1_9: In HEVC encoding, the PU partition mode is determined by RDO technology. In order to accurately measure the bit rate increase and picture quality degradation caused by modifying the PU partition mode, the present invention adopts the RD cost (RD cost) of the improved RDO as part of the distortion cost, that is, to assign weight control factors to the distortion and the number of bits respectively, so that it favors the carrier with less impact on the embedding bit rate and reasonable visual distortion during the embedding process. Calculate the cost value of the PU partition mode by modifying the carrier corresponding to each integer in the first type of carrier sequence to the carrier corresponding to other integers in the same type of carrier sequence. For the carrier corresponding to any integer in the first type of carrier sequence, take it as the current carrier, set the current carrier to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and modify the current carrier to the carrier corresponding to any other integer t in the same type of carrier sequence, that is, the cost value of the PU partition mode is recorded as

Calculate the cost of changing the carrier corresponding to each integer in the second type of carrier sequence to the carrier corresponding to other integers in the same type of carrier sequence, that is, the PU partition mode. For any carrier corresponding to an integer in the second type of carrier sequence, use it as the current carrier, set the current carrier to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and change the current carrier to the carrier corresponding to any other integer s in the same type of carrier sequence, that is, the PU partition mode. The cost value is recorded as

表示将当前载体修改为同类载体序列中的其它任一个整数t对应的载体即PU划分模式后的失真，

表示将当前载体修改为同类载体序列中的其它任一个整数t对应的载体即PU划分模式后的比特数，

表示将当前载体修改为同类载体序列中的其它任一个整数t对应的载体即PU划分模式后的拉格朗日因子，

表示失真控制因子，

表示比特率控制因子，

经实验测试发现

时视频具有较低的比特率增长且视频质量影响较低，嵌入效果较好，若当前载体属于第二类载体，则当前载体对应的整数为0时s为1或2，当前载体对应的整数为1时s为0或2，当前载体对应的整数为2时s为0或1，即当前载体对应有2个代价值，

表示将当前载体修改为同类载体序列中的其它任一个整数s对应的载体即PU划分模式后的失真，

表示将当前载体修改为同类载体序列中的其它任一个整数s对应的载体即PU划分模式后的比特数，

表示将当前载体修改为同类载体序列中的其它任一个整数s对应的载体即PU划分模式后的拉格朗日因子，

和

均为改进RDO的RD cost。Wherein, 1≤m≤M, M represents the total number of frames contained in the H.265/HEVC video, 1≤n≤N, N represents the total number of coding tree units contained in the m-th frame in the H.265/HEVC video, 1≤n≤N, 1≤k≤K, K represents the total number of coding units contained in the n-th coding tree unit in the m-th frame in the H.265/HEVC video, and if the current carrier belongs to the first type of carrier, when the integer corresponding to the current carrier is 0, t is 1, 2, 3, 4 or 5, when the integer corresponding to the current carrier is 1, t is 0, 2, 3, 4 or 5, when the integer corresponding to the current carrier is 2, t is 0, 1, 3, 4 or 5, when the integer corresponding to the current carrier is 3, t is 0, 1, 2, 4 or 5, and when the integer corresponding to the current carrier is 4 t is 0, 1, 2, 3 or 5. When the integer corresponding to the current carrier is 5, t is 0, 1, 2, 3 or 4, that is, the current carrier corresponds to 5 cost values. exp() represents an exponential function with the natural base e as the base, e=2.71…, α represents the inter-frame distortion transmission rate, 0＜α＜1, and in this embodiment, α is 0.5. GOPSize represents the total number of frames in a picture group GOP in the H.265/HEVC video. In this embodiment, the value of GOPSize is 4. Poc represents the position of the frame where the coding unit corresponding to the current carrier is located in the picture group GOP where the frame is located, that is, the frame where the coding unit corresponding to the current carrier is located is the Poc-th frame in the picture group GOP where the frame is located, 1≤Poc≤GOPSize,

It indicates the distortion after changing the current carrier to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

Indicates the number of bits after the current carrier is modified to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

It indicates the Lagrangian factor after the current carrier is modified to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

represents the distortion control factor,

represents the bit rate control factor,

Experimental tests found that

When the video has a lower bit rate growth and a lower impact on video quality, the embedding effect is better. If the current carrier belongs to the second type of carrier, when the integer corresponding to the current carrier is 0, s is 1 or 2, when the integer corresponding to the current carrier is 1, s is 0 or 2, and when the integer corresponding to the current carrier is 2, s is 0 or 1, that is, the current carrier corresponds to 2 cost values.

It indicates the distortion after changing the current carrier to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode.

Indicates the number of bits after the current carrier is modified to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode.

It indicates the Lagrangian factor after the current carrier is modified to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode.

and

Both are RD costs of improved RDO.

取值为0.3，

取值为0.7。In this embodiment, in step 1_9, α is set to 0.5.

The value is 0.3.

The value is 0.7.

和

的值取决于是否采用Merge帧间预测模式，当采用Merge帧间预测模式时，

和

的值为SATD(Sum of AbsoluteTransformed Difference)值；当采用非Merge帧间预测模式(AMVP技术)时，

和

的值为编码失真。在采用HM编码器对HEVC视频进行压缩编码过程中，将AMP_ENC_SPEEDUP快速模式关闭，即令AMP_ENC_SPEEDUP＝0即可直接获取各PU划分模式对应的失真。In this embodiment, in steps 1-9,

and

The value of depends on whether the Merge inter-frame prediction mode is used. When the Merge inter-frame prediction mode is used,

and

The value is the SATD (Sum of Absolute Transformed Difference) value; when the non-Merge inter-frame prediction mode (AMVP technology) is used,

and

The value of is the coding distortion. When the HM encoder is used to compress and encode the HEVC video, the AMP_ENC_SPEEDUP fast mode is turned off, that is, AMP_ENC_SPEEDUP=0, and the distortion corresponding to each PU division mode can be directly obtained.

和

的值取决于是否采用Merge帧间预测模式，当采用Merge帧间预测模式时，

和

的值为Merge索引的编码比特数；当采用非Merge帧间预测模式(AMVP技术)时，

和

的值为表示预测模式和残差变换系数的编码比特数。在采用HM编码器对HEVC视频进行压缩编码过程中，将AMP_ENC_SPEEDUP快速模式关闭，即令AMP_ENC_SPEEDUP＝0即可直接获取各PU划分模式对应的比特数。In this embodiment, in steps 1-9,

and

The value of depends on whether the Merge inter-frame prediction mode is used. When the Merge inter-frame prediction mode is used,

and

The value is the number of coded bits of the Merge index; when the non-Merge inter-frame prediction mode (AMVP technology) is used,

and

The value of represents the number of coding bits for the prediction mode and the residual transform coefficient. In the process of compressing and encoding the HEVC video using the HM encoder, the AMP_ENC_SPEEDUP fast mode is turned off, that is, AMP_ENC_SPEEDUP=0, and the number of bits corresponding to each PU partition mode can be directly obtained.

和

的值通过μ·ω·2^(QP-12)/3.0计算得到，其中，μ的取值依赖于当前载体对应的编码单元所在的帧是否为该帧所在的图像组GOP内的参考帧，若该帧为其所在的图像组GOP内的参考帧，则μ＝1.0-Clip3(0.0,0.5,0.05N_B)，若该帧为其所在的图像组GOP内的非参考帧，则μ＝1.0，N_B表示H.265/HEVC视频中一个图像组GOP内的B帧的总帧数(LDP配置中P帧在HM中也算作B帧)，Clip3()为Clip操作函数，ω表示加权因子，ω的值由编码配置和当前载体对应的编码单元所在的帧在该帧所在的图像组GOP内的位置Poc决定，QP表示当前载体对应的编码单元所在的帧的编码量化参数。在采用HM编码器对HEVC视频进行压缩编码过程中，将AMP_ENC_SPEEDUP快速模式关闭，即令AMP_ENC_SPEEDUP＝0即可直接获取各PU划分模式对应的拉格朗日因子。In this embodiment, in steps 1-9,

and

The value of is calculated by μ·ω·2 ^(QP-12)/3.0 , where the value of μ depends on whether the frame where the coding unit corresponding to the current carrier is located is a reference frame in the picture group GOP where the frame is located. If the frame is a reference frame in the picture group GOP where the frame is located, then μ=1.0-Clip3( _{0.0,0.5,0.05NB} ); if the frame is a non-reference frame in the picture group GOP where the frame is located, then μ=1.0, _NB represents the total number of B frames in a picture group GOP in H.265/HEVC video (P frames in LDP configuration are also counted as B frames in HM), Clip3() is the Clip operation function, ω represents the weighting factor, and the value of ω is determined by the encoding configuration and the position Poc of the frame where the coding unit corresponding to the current carrier is located in the picture group GOP where the frame is located, and QP represents the encoding quantization parameter of the frame where the coding unit corresponding to the current carrier is located. In the process of compressing and encoding the HEVC video using the HM encoder, the AMP_ENC_SPEEDUP fast mode is turned off, that is, AMP_ENC_SPEEDUP=0, so that the Lagrangian factors corresponding to each PU partitioning mode can be directly obtained.

将当前整数保持不变时的失真代价值记为

将当前整数减1时的失真代价值记为

若当前整数z_m,n,k为0，则

为无穷大；若当前整数z_m,n,k为1，则

若当前整数z_m,n,k为2，则

为无穷大、

若当前整数z_m,n,k为3，则

为无穷大；若当前整数z_m,n,k为4，则

若当前整数z_m,n,k为5，则

为无穷大、

为了防止PU划分模式大幅度修改导致对视频内容的不准确识别，在此规定PU划分模式的修改在同方向内进行，即水平划分模式与垂直划分模式之间不能相互修改。Step 1_10: Calculate the distortion cost value when each integer in the first type of carrier sequence is increased by 1, remains unchanged, or decreased by 1. For any integer in the first type of carrier sequence, take it as the current integer, set the carrier corresponding to the current integer to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and record the distortion cost value when the current integer is increased by 1 as

The distortion cost when the current integer remains unchanged is recorded as

The distortion cost when the current integer is reduced by 1 is recorded as

If the current integer z _m,n,k is 0, then

is infinite; if the current integer z _m,n,k is 1, then

If the current integer z _m,n,k is 2, then

For infinity,

If the current integer z _m,n,k is 3, then

is infinite; if the current integer z _m,n,k is 4, then

If the current integer z _m,n,k is 5, then

For infinity,

In order to prevent the PU partition mode from being modified significantly and causing inaccurate recognition of the video content, it is stipulated here that the modification of the PU partition mode is performed in the same direction, that is, the horizontal partition mode and the vertical partition mode cannot be modified mutually.

将当前整数保持不变时的失真代价值记为

将当前整数减1时的失真代价值记为

若当前整数

为0，则

为无穷大；若当前整数

为1，则

若当前整数

为2，则

为无穷大、

Calculate the distortion cost value when each integer in the second type of carrier sequence is increased by 1, remains unchanged, or decreased by 1. For any integer in the second type of carrier sequence, take it as the current integer, set the carrier corresponding to the current integer to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and record the distortion cost value when the current integer is increased by 1 as

The distortion cost when the current integer remains unchanged is recorded as

The distortion cost when the current integer is reduced by 1 is recorded as

If the current integer

If is 0,

is infinite; if the current integer

If is 1,

If the current integer

is 2, then

For infinity,

Step 1_11: Use the same embedding load rate payload to randomly generate two different binary secret information sequences, which are denoted as _X1 and _X2 respectively; then, according to the distortion cost value when each integer in the first type of carrier sequence is increased by 1, remains unchanged, or decreased by 1, use the STC (Syndrome-Trellis Codes) toolkit to embed _X1 into all integers in the first type of carrier sequence to obtain the first type of secret carrier sequence; similarly, according to the distortion cost value when each integer in the second type of carrier sequence is increased by 1, remains unchanged, or decreased by 1, use the STC toolkit to embed _X2 into all integers in the second type of carrier sequence to obtain the second type of secret carrier sequence; wherein, payload∈(0,0.5], in this embodiment, the value of payload is 0.5, the length of _X1 is _Num1 ×payload, the length of _X2 is _Num2 ×payload, _Num1 represents the total number of carriers in the first type of carrier, and _Num2 represents the total number of carriers in the second type of carrier.

Step 1_12: Convert all integers in the first type of secret carrier sequence and the second type of secret carrier sequence into PU partition patterns respectively. The specific process is as follows:

There are six integers in the first type of secret carrier sequence, namely 0, 1, 2, 3, 4, and 5. The integer 0 is converted into a 2N×N partitioning pattern, the integer 1 is converted into a 2N×nU partitioning pattern, the integer 2 is converted into a 2N×nD partitioning pattern, the integer 3 is converted into an N×2N partitioning pattern, the integer 4 is converted into an nL×2N partitioning pattern, and the integer 5 is converted into an nR×2N partitioning pattern. Then all the PU partitioning patterns obtained after converting all the integers in the first type of secret carrier sequence are used to replace the carriers in the same position in the first type of carriers in sequence to form the first type of secret carriers.

The second type of secret carrier sequence has three integers, namely 0, 1, 2, and 3. The integer 0 is converted into a 2N×N partition pattern, the integer 1 is converted into an N×2N partition pattern, and the integer 2 is converted into an N×N partition pattern. Then all the PU partition patterns obtained after converting all the integers in the second type of secret carrier sequence are used to replace the carriers in the same position in the second type of carriers in sequence to form the second type of secret carriers.

Step 1_13: Use the H.265/HEVC standard encoder to compress and encode the original video. In the prediction process of compression coding, the original PU division pattern at the corresponding position is replaced by the steganographic first-type secret carrier and the second-type secret carrier, and the secret video code stream is obtained through compression coding.

The specific process of extracting secret information is as follows:

Step 2_1: Use the H.265/HEVC standard decoder to decode the encrypted video stream. During the decoding process, save the prediction mode and depth of each coding unit in each coding tree unit in each frame, and the PU partition mode corresponding to each coding unit, and obtain the decoded video after the decoding is completed.

Step 2_2: According to the process of obtaining the first type of carrier sequence and the second type of carrier sequence from step 1_2 to step 1_8, the encrypted first type of carrier sequence and the encrypted second type of carrier sequence corresponding to the decoded video are obtained in the same manner.

同样，利用STC工具包对步骤2_2得到的含密的第二类载体序列进行解码，提取得到第二隐秘信息，记为

其中，

Ext_STC()表示STC解码，y₁表示含密的第一类载体序列，H表示STC的奇偶校验矩阵，在发送者和接收者间共享，y₂表示含密的第二类载体序列。Step 2_3: Use the STC toolkit to decode the first type of carrier sequence obtained in step 2_2, and extract the first secret information, which is recorded as

Similarly, the STC toolkit is used to decode the secret second type carrier sequence obtained in step 2_2, and the second secret information is extracted, which is recorded as

in,

Ext _STC () represents STC decoding, y ₁ represents the first type of carrier sequence containing secrets, H represents the parity check matrix of STC, which is shared between the sender and the receiver, and y ₂ represents the second type of carrier sequence containing secrets.

In order to further illustrate the feasibility and effectiveness of the method of the present invention, experiments were conducted on the method of the present invention.

The experiment selected several representative video sequences with a resolution of 1920×1080, namely BasketballDrive, BQTerrace, Cactus, and ParkScene. The method of the present invention is carried out under the H.265/HEVC reference software HM16.15, and the built-in encoder_lowdelay_P_main.cfg configuration file is used for encoding, wherein the encoding structure of the video sequence is IPPP, the length of a group of images GOP is 4, and a group of images GOP contains 20 frames (that is, the first 20 frames of each video sequence are selected), the encoding quantization parameter QP is fixed to 32, and the other encoding parameters are consistent with the original configuration file.

In order to further illustrate the effectiveness of the method of the present invention, five indicators, namely Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM), Mean Squared Error (MSE), Bit Rate Increase (BRI) and steganalysis detection rate, are used to evaluate the steganographic effect of HEVC video data. The larger the PSNR and SSIM values, the better the video picture quality; the smaller the MSE, the better the video picture quality; the smaller the BRI, the smaller the impact of the steganography method on the video bit rate; the closer the steganalysis detection rate is to 50%, the more the steganography method can deceive steganalysis detection.

Table 1 shows the comparison of the PSNR, SSIM, and MSE results of the video obtained by compressing the original video using only the HM encoder, the encrypted video obtained by steganographically writing the original video using the Zhang method, and the encrypted video obtained by steganographically writing the original video using the method of the present invention.

Table 1 Comparison of PSNR, SSIM, and MSE obtained by three methods

In Table 1, “Original” refers to the method of compressing the original video using the HM encoder, and Zhang’s method refers to Z. Zhang, Z. Li, J. Liu, H. Yan, and L. Yu, Steganography algorithm based on modified EMD-coded PU partition modes for HEVC videos[J], EURASIP Journal on Image and Video Processing, 2021(1): 1–20. (A HEVC video steganography method based on improved EMD-coded PU partition mode).

It can be concluded from the data listed in Table 1 that the encrypted video sequence obtained by the method of the present invention is superior to the comparative method in various evaluation indicators of video picture quality.

Figure 2 shows the results of the encrypted video obtained by using Zhang's method to steganographically write data on the original video and the encrypted video obtained by using the proposed method to hide data on the original video on BRI. From the data shown in Figure 2, it can be concluded that the bit rate increase of the encrypted video sequence obtained by the proposed method is small, and its bit rate impact is smaller than that of the comparison method, indicating that the proposed method has an excellent effect of suppressing the bit rate increase.

Table 2 shows the results of the anti-steganographic analysis of the secret video obtained by using Zhang's method to steganographic the original video and the secret video obtained by using the method of the present invention to steganographic the original video. The Zhai method refers to L. Zhai, L. Wang, and Y. Ren, Universal detection of video steganography in multiple domains based on the consistency of motion vectors [J], IEEE Transactions on Information Forensics and Security, vol. 15, pp. 1762–1777, 2019. (A universal detection method for multi-domain video steganography based on the consistency of motion vectors).

Table 2 Comparison of the anti-steganalysis performance of the two methods

Steganalysis Detection Methods Method of the present invention Zhang Method Zhai Method 50% 94%

It can be concluded from the data listed in Table 2 that the anti-steganalysis performance of the encrypted video sequence obtained by the method of the present invention is higher than that of the comparison method, indicating that the security of the method of the present invention is higher.

Claims (5)

1. An H.265/HEVC video adaptive steganography method based on improved RDO, including two parts: secret information embedding and secret information extraction; the characteristics are: The specific process of embedding the secret information is as follows: Step 1_1: Use the H.265/HEVC standard encoder to compress and encode the original video. During the compression and encoding process, the prediction mode and depth of each coding unit in each coding tree unit in each frame and the PU partition mode corresponding to each coding unit are saved. After the compression and encoding, the H.265/HEVC video is obtained; wherein the size of the coding tree unit is 64×64, and the size of the coding unit is 64×64, 32×32, 16×16, or 8×8; Step 1_2: traverse all frames in the H.265/HEVC video in order, and define the currently traversed frame as the current frame; Step 1_3: Determine whether the current frame is a P frame. If the current frame is a P frame, execute step 1_4; if the current frame is not a P frame, directly execute step 1_7; Step 1_4: traverse the coding units of all sizes in the current frame in order, and define the currently traversed coding unit as the current unit; Step 1_5: If the size of the current unit is 64×64, the current unit is not processed, and then step 1_6 is executed; if the size of the current unit is 32×32 or 16×16, the PU partition mode corresponding to the current unit is used as a carrier and classified as a first-class carrier, and the index position of the current unit in the H.265/HEVC video, the depth of the current unit, and the PU partition mode corresponding to the current unit are recorded, and then step 1_6 is executed; if the size of the current unit is 8×8, the PU partition mode corresponding to the current unit is used as a carrier and classified as a second-class carrier, and the index position of the current unit in the H.265/HEVC video, the depth of the current unit, and the PU partition mode corresponding to the current unit are recorded, and then step 1_6 is executed; Step 1_6: traverse the next coding unit in the current frame as the current unit, then return to step 1_5 to continue executing until all coding units in the current frame are traversed, and then execute step 1_7; Step 1_7: traverse the next frame in the H.265/HEVC video as the current frame, then return to step 1_3 to continue executing, and execute step 1_8 after all frames in the H.265/HEVC video have been traversed; Step 1_8: Map all carriers classified as the first type of carriers and all carriers classified as the second type of carriers, that is, PU partitioning modes, respectively. The specific process is as follows: There are eight PU partition modes for the first type of carriers, namely 2N×2N, N×N, 2N×N, 2N×nU, 2N×nD, N×2N, nL×2N, and nR×2N partition modes. The 2N×N, 2N×nU, and 2N×nD partition modes are all horizontal partition modes, and the N×2N, nL×2N, and nR×2N partition modes are all vertical partition modes. For any carrier classified as the first type of carrier, if the carrier is in the 2N×2N partition mode or the N×N partition mode, the carrier is not mapped; if the carrier is in the 2N×N partition mode, the carrier is mapped to the integer 0; if the carrier is in the 2N×nU partition mode, the carrier is mapped to the integer 0; If the carrier is a 2N×nD partition mode, the carrier is mapped to an integer 1; if the carrier is a 2N×nD partition mode, the carrier is mapped to an integer 2; if the carrier is an N×2N partition mode, the carrier is mapped to an integer 3; if the carrier is an nL×2N partition mode, the carrier is mapped to an integer 4; if the carrier is an nR×2N partition mode, the carrier is mapped to an integer 5; then all integers obtained after mapping all carriers classified as the first type of carriers are arranged in the order of the index positions of the coding units corresponding to the carrier, i.e., the PU partition mode, in the H.265/HEVC video to form a first type of carrier sequence; then execute step 1_9; There are four PU partition modes for the second type of carrier, namely 2N×2N, 2N×N, N×2N, and N×N partition modes; for any carrier classified as the second type of carrier, if the carrier is in the 2N×2N partition mode, the carrier is not mapped; if the carrier is in the 2N×N partition mode, the carrier is mapped to an integer 0; if the carrier is in the N×2N partition mode, the carrier is mapped to an integer 1; if the carrier is in the N×N partition mode, the carrier is mapped to an integer 2; then all integers obtained after mapping all carriers classified as the second type of carrier are arranged in the order of the index positions of the coding units corresponding to the carrier, i.e., the PU partition mode, in the H.265/HEVC video to form a second type of carrier sequence; then execute step 1_9; Step 1_9: Calculate the cost of modifying the carrier corresponding to each integer in the first type of carrier sequence to the carrier corresponding to other integers in the same type of carrier sequence, that is, the PU partition mode. For any carrier corresponding to an integer in the first type of carrier sequence, use it as the current carrier, set the current carrier to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and modify the current carrier to the carrier corresponding to any other integer t in the same type of carrier sequence, that is, the PU partition mode. The cost value is recorded as

Calculate the cost of modifying the carrier corresponding to each integer in the second type of carrier sequence to the carrier corresponding to other integers in the same type of carrier sequence, that is, the PU partition mode. For any carrier corresponding to an integer in the second type of carrier sequence, use it as the current carrier, set the current carrier to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and modify the current carrier to the carrier corresponding to any other integer s in the same type of carrier sequence, that is, the PU partition mode. The cost value is recorded as

Wherein, 1≤m≤M, M represents the total number of frames contained in the H.265/HEVC video, 1≤n≤N, N represents the total number of coding tree units contained in the m-th frame in the H.265/HEVC video, 1≤n≤N, 1≤k≤K, K represents the total number of coding units contained in the n-th coding tree unit in the m-th frame in the H.265/HEVC video, if the current carrier belongs to the first type of carrier, then when the integer corresponding to the current carrier is 0, t is 1, 2, 3, 4 or 5, when the integer corresponding to the current carrier is 1, t is 0, 2, 3, 4 or 5, when the integer corresponding to the current carrier is 2, t is 0, 1, 3, 4 or 5, and the integer corresponding to the current carrier is When the integer corresponding to the current carrier is 3, t is 0, 1, 2, 4 or 5. When the integer corresponding to the current carrier is 4, t is 0, 1, 2, 3 or 5. When the integer corresponding to the current carrier is 5, t is 0, 1, 2, 3 or 4. exp() represents an exponential function with the natural base e as the base. α represents the inter-frame distortion transmission rate, 0＜α＜1. GOPSize represents the total number of frames in a picture group GOP in H.265/HEVC video. Poc represents the position of the frame where the coding unit corresponding to the current carrier is located in the picture group GOP where the frame is located. That is, the frame where the coding unit corresponding to the current carrier is located is the Poc-th frame in the picture group GOP where the frame is located. 1≤Poc≤GOPSize.

It indicates the distortion after changing the current carrier to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

Indicates the number of bits after the current carrier is modified to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

It indicates the Lagrangian factor after the current carrier is modified to the carrier corresponding to any other integer t in the same carrier sequence, that is, the PU partition mode.

represents the distortion control factor,

represents the bit rate control factor,

If the current carrier belongs to the second type of carrier, then when the integer corresponding to the current carrier is 0, s is 1 or 2; when the integer corresponding to the current carrier is 1, s is 0 or 2; when the integer corresponding to the current carrier is 2, s is 0 or 1.

It indicates the distortion after changing the current carrier to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode.

Indicates the number of bits after the current carrier is modified to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode.

Indicates the Lagrangian factor after the current carrier is modified to the carrier corresponding to any other integer s in the same carrier sequence, that is, the PU partition mode; Step 1_10: Calculate the distortion cost value when each integer in the first type of carrier sequence is increased by 1, remains unchanged, or decreased by 1. For any integer in the first type of carrier sequence, take it as the current integer, set the carrier corresponding to the current integer to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and record the distortion cost value when the current integer is increased by 1 as

The distortion cost when the current integer remains unchanged is recorded as

The distortion cost when the current integer is reduced by 1 is recorded as

If the current integer z _m,n,k is 0, then

is infinite; if the current integer z _m,n,k is 1, then

If the current integer z _m,n,k is 2, then

For infinity,

If the current integer z _m,n,k is 3, then

is infinite; if the current integer z _m,n,k is 4, then

If the current integer z _m,n,k is 5, then

For infinity,

Calculate the distortion cost value when each integer in the second type of carrier sequence is increased by 1, remains unchanged, or decreased by 1. For any integer in the second type of carrier sequence, take it as the current integer, set the carrier corresponding to the current integer to the PU partition mode corresponding to the kth coding unit in the nth coding tree unit in the mth frame of the H.265/HEVC video, and record the distortion cost value when the current integer is increased by 1 as

The distortion cost when the current integer remains unchanged is recorded as

The distortion cost when the current integer is reduced by 1 is recorded as

If the current integer

If is 0,

is infinite; if the current integer

is 1, then

If the current integer

is 2, then

For infinity,

Step 1_11: Use the same embedding load rate payload to randomly generate two different binary secret information sequences, which are denoted as _X1 and _X2 respectively; then, according to the distortion cost value when each integer in the first type of carrier sequence is increased by 1, remains unchanged, or decreased by 1, use the STC toolkit to embed _X1 into all integers in the first type of carrier sequence to obtain the first type of secret carrier sequence; similarly, according to the distortion cost value when each integer in the second type of carrier sequence is increased by 1, remains unchanged, or decreased by 1, use the STC toolkit to embed _X2 into all integers in the second type of carrier sequence to obtain the second type of secret carrier sequence; where payload∈(0,0.5], the length of _X1 is _Num1 ×payload, the length of _X2 is _Num2 ×payload, _Num1 represents the total number of carriers in the first type of carrier, and _Num2 represents the total number of carriers in the second type of carrier; Step 1_12: Convert all integers in the first type of secret carrier sequence and the second type of secret carrier sequence into PU partition patterns respectively. The specific process is as follows: The first type of secret carrier sequence has six integers, namely 0, 1, 2, 3, 4, and 5. The integer 0 is converted into a 2N×N partitioning mode, the integer 1 is converted into a 2N×nU partitioning mode, the integer 2 is converted into a 2N×nD partitioning mode, the integer 3 is converted into an N×2N partitioning mode, the integer 4 is converted into an nL×2N partitioning mode, and the integer 5 is converted into an nR×2N partitioning mode. Then, all the PU partitioning modes obtained after converting all the integers in the first type of secret carrier sequence are used to replace the carriers in the same position in the first type of carriers in order to form the first type of secret carriers. The second type of secret carrier sequence has three integers, namely 0, 1, 2, and 3. The integer 0 is converted into a 2N×N partition mode, the integer 1 is converted into an N×2N partition mode, and the integer 2 is converted into an N×N partition mode. Then, all the PU partition modes obtained after converting all the integers in the second type of secret carrier sequence are used to replace the carriers at the same position in the second type of carriers in order to form the second type of secret carriers. Step 1_13: Use the H.265/HEVC standard encoder to compress and encode the original video. In the prediction process of compression encoding, the original PU partitioning pattern at the corresponding position is replaced with the first type of secret carrier and the second type of secret carrier after steganography, and the secret video code stream is obtained through compression encoding; The specific process of extracting secret information is as follows: Step 2_1: Use the H.265/HEVC standard decoder to decode the encrypted video stream. During the decoding process, save the prediction mode and depth of each coding unit in each coding tree unit in each frame, and the PU partition mode corresponding to each coding unit, and obtain the decoded video after decoding; Step 2_2: Obtain the encrypted first-type carrier sequence and the encrypted second-type carrier sequence corresponding to the decoded video in the same manner as the process of obtaining the first-type carrier sequence and the second-type carrier sequence in step 1_2 to step 1_8; Step 2_3: Use the STC toolkit to decode the first type of carrier sequence obtained in step 2_2, and extract the first secret information, which is recorded as

Similarly, the STC toolkit is used to decode the secret second type carrier sequence obtained in step 2_2, and the second secret information is extracted, which is recorded as

2. According to the H.265/HEVC video adaptive steganography method based on improved RDO according to claim 1, it is characterized in that in the steps 1_9, the value of α is 0.5,

The value is 0.3.

The value is 0.7. 3. The H.265/HEVC video adaptive steganography method based on improved RDO according to claim 1 or 2, characterized in that in the steps 1-9,

and

The value of depends on whether the Merge inter-frame prediction mode is used. When the Merge inter-frame prediction mode is used,

and

The value is the SATD value; when the non-Merge inter-frame prediction mode is used,

and

The value of is the coding distortion. 4. The H.265/HEVC video adaptive steganography method based on improved RDO according to claim 3, characterized in that in the steps 1-9,

and

The value of depends on whether the Merge inter-frame prediction mode is used. When the Merge inter-frame prediction mode is used,

and

The value is the number of coded bits of the Merge index; when the non-Merge inter-frame prediction mode is used,

and

The value represents the number of coding bits for the prediction mode and the residual transform coefficient. 5. The H.265/HEVC video adaptive steganography method based on improved RDO according to claim 4, characterized in that in the steps 1-9,

and

The value of is calculated by μ·ω·2 ^(QP-12)/3.0 , where the value of μ depends on whether the frame where the coding unit corresponding to the current carrier is located is a reference frame in the picture group GOP where the frame is located. If the frame is a reference frame in the picture group GOP where the frame is located, then μ＝1.0-Clip3( _{0.0,0.5,0.05NB} ); if the frame is a non-reference frame in the picture group GOP where the frame is located, then μ＝1.0, _NB represents the total number of B frames in a picture group GOP in the H.265/HEVC video, Clip3() is the Clip operation function, ω represents the weighting factor, and the value of ω is determined by the encoding configuration and the position Poc of the frame where the coding unit corresponding to the current carrier is located in the picture group GOP where the frame is located, and QP represents the encoding quantization parameter of the frame where the coding unit corresponding to the current carrier is located.

Images