CN114727112B - H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization - Google Patents

H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization Download PDF

Info

Publication number
CN114727112B
CN114727112B CN202210172234.9A CN202210172234A CN114727112B CN 114727112 B CN114727112 B CN 114727112B CN 202210172234 A CN202210172234 A CN 202210172234A CN 114727112 B CN114727112 B CN 114727112B
Authority
CN
China
Prior art keywords
qdst
carrier
sequence
coefficient
unit
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202210172234.9A
Other languages
Chinese (zh)
Other versions
CN114727112A (en
Inventor
王让定
杨璘
徐达文
董理
何松翰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ningbo University
Ningbo University of Technology
Original Assignee
Ningbo University
Ningbo University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ningbo University, Ningbo University of Technology filed Critical Ningbo University
Priority to CN202210172234.9A priority Critical patent/CN114727112B/en
Publication of CN114727112A publication Critical patent/CN114727112A/en
Application granted granted Critical
Publication of CN114727112B publication Critical patent/CN114727112B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/46—Embedding additional information in the video signal during the compression process
    • H04N19/467—Embedding additional information in the video signal during the compression process characterised by the embedded information being invisible, e.g. watermarking
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
    • H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
    • H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
    • H04N19/96—Tree coding, e.g. quad-tree coding

Abstract

The invention discloses a distortion compensation optimization-based H.265/HEVC video self-adaptive information hiding method, wherein secret information embedding comprises three parts of extracting carriers, calculating distortion cost and distortion compensation, extracting proper carriers from QDST coefficients in 4X 4 coding units in all stored I frames according to a threshold value, calculating the distortion cost of the carriers during simulation embedding, realizing a data hiding mode of realizing the minimum total distortion cost by using an STC algorithm for the carriers according to the distortion cost, and carrying out distortion compensation on coefficient groups corresponding to the carriers according to difference values before and after steganography to eliminate intra-frame distortion drift caused by modulating the QDST coefficients; because the method comprehensively considers factors such as video picture quality, video block texture complexity, inter-frame distortion transmission, code stream and the like when calculating the distortion cost value, the secret-containing video steganographically by the method has better picture quality, lower code rate increase and higher safety.

Description

H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization
Technical Field
The invention relates to a video information hiding technology, in particular to an H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization.
Background
In today's society, intelligent devices have become more and more popular, and people can obtain various information contents through text, audio, image or video channels. However, at the same time, the phenomenon that the digital information is tampered, imitated or attacked is becoming serious, and how to protect the information content of the digital media and to suppress the illegal use of the digital media has become a serious problem to be solved. Information hiding is a technology for hiding secret information in a public transmission channel and cannot be perceived by people in visual sense and hearing sense, and is a powerful secret communication and copyright protection means. Among digital multimedia information hiding technologies, an adaptive information hiding technology is an extremely important component, which can adaptively embed hidden information into a carrier according to a distortion function, and can minimize embedded distortion through steganographic encoding such as STC (syncrome-Trellis Codes) and the like, thereby achieving excellent visual effects and efficient encoding efficiency.
Because the data volume of digital video is generally huge, the digital video is usually transmitted in a compressed form in a public transmission channel, HEVC (High Efficiency Video Coding ) is the latest generation video compression coding standard, and the core aim is to improve the compression efficiency by one time on the basis of H.264/AVC, thereby playing an important role in the video application fields of high definition, ultra-high definition and the like. Therefore, the research on the self-adaptive information hiding technology of the H.265/HEVC video has great practical significance and theoretical value. Although adaptive information hiding technology has many researches in the field of audio and image, because of the difference between video carriers and both audio and image, these researches cannot be directly migrated into the field of video information hiding technology, and thus, it is necessary to study the characteristics of h.265/HEVC video and design a suitable and efficient adaptive distortion function for its video characteristics to reduce embedded distortion.
Disclosure of Invention
The technical problem to be solved by the invention is to provide the H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization, which can well improve video visual quality, has less influence on video code stream modification and has high safety.
The technical scheme adopted for solving the technical problems is as follows: an H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization comprises two parts of secret information embedding and secret information extracting; the method is characterized in that:
the specific process of the secret information embedding is as follows:
step 1_1: performing compression coding on an original video by adopting an H.265/HEVC standard coder to obtain an H.265/HEVC video, and storing a prediction mode and depth of each coding unit in each coding tree unit in each frame in the H.265/HEVC video, a PU partition mode corresponding to each coding unit and a QDST coefficient of each pixel point in each coding unit in the compression coding process; 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 or 4×4;
step 1_2: traversing all frames in the H.265/HEVC video in sequence, and defining the currently traversed frame as a current frame;
step 1_3: judging whether the current frame is an I frame or not, and if the current frame is the I frame, executing the step 1_4; if the current frame is not the I frame, directly executing the step 1_8;
step 1_4: traversing all coding units with the size of 4 multiplied by 4 in the current frame in sequence, and defining the currently traversed coding units with the size of 4 multiplied by 4 as current units;
Step 1_5: classifying the current unit as a first type candidate block, a second type candidate block or a third type candidate block, or not processing the current unit, wherein the specific process is as follows:
if the current unit has only adjacent 4×4 lower left coding units and the number of prediction modes of the lower left coding units is identified in the interval [2,26], and adjacent 4×4 lower right coding units and the number of prediction modes of the lower right coding units is identified in the interval [2,10], classifying the current unit as a first type candidate block, and then performing step 1_6;
if the current unit has only the adjacent upper right coding unit of size 4×4 and the number of the prediction mode of the upper right coding unit is identified in the interval [10,34] or 1, and the adjacent right coding unit of size 4×4 and the number of the prediction mode of the right coding unit is identified in the interval [26,34], classifying the current unit as a second type candidate block, and then performing step 1_6;
if the current unit has adjacent 4×4 lower left coding unit and the number of prediction modes of the lower left coding unit is within the interval [2,26], and adjacent 4×4 lower right coding unit and the number of prediction modes of the lower right coding unit is within the interval [2,10], and adjacent 4×4 upper right coding unit and the number of prediction modes of the upper right coding unit is within the interval [10,34] or 1, and adjacent 4×4 right coding unit and the number of prediction modes of the lower right coding unit is within the interval [26,34], and adjacent 4×4 lower right coding unit and the number of prediction modes of the lower right coding unit is within the interval [26,34] or 0 or 1, classifying the current unit as a third type candidate block, and then performing step 1_6;
For the rest of the cases, the current unit is not processed, and then step 1_7 is directly performed;
the above, the lower left coding unit is a coding unit with a size of 4×4 located right below the current unit, the lower right coding unit is a coding unit with a size of 4×4 located right below the current unit, the upper right coding unit is a coding unit with a size of 4×4 located right above the current unit, the lower right coding unit is a coding unit with a size of 4×4 located right above the current unit, and the lower right coding unit is a coding unit with a size of 4×4 located right below the current unit;
step 1_6: selecting carriers from the QDST coefficients of all pixel points in the current unit, if the current unit is a first type candidate block, selecting QDST coefficients with absolute values larger than or equal to a set threshold th from 4 QDST coefficients of the 1 st column pixel point of the current unit as carriers, classifying the carriers as first type carriers, and then executing step 1_7; if the current unit is a second type candidate block, selecting QDST coefficients with absolute values larger than or equal to a set threshold th from 4 QDST coefficients of a 1 st row pixel point of the current unit as carriers, classifying the carriers as a second type carrier, and then executing step 1_7; if the current unit is a third type candidate block, selecting QDST coefficients with absolute values larger than or equal to a set threshold th from 16 QDST coefficients of all pixel points of the current unit as carriers, classifying the carriers into a third type carrier, and then executing step 1_7; wherein th E (0, 10);
Step 1_7: traversing the next 4×4 coding unit in the current frame as the current unit, returning to step 1_5, continuing to execute until all the 4×4 coding units in the current frame are traversed, and then executing step 1_8;
step 1_8: traversing the next frame in the H.265/HEVC video as the current frame, and returning to the step 1_3 to continue execution until all frames in the H.265/HEVC video are traversed completely;
step 1_9: arranging all carriers classified as the first type of carriers, namely QDST coefficients, according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a first type of carrier sequence; likewise, all carriers classified as the second type of carriers, namely QDST coefficients, are arranged according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a second type of carrier sequence; arranging all carriers classified as third-class carriers, namely QDST coefficients, according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a third-class carrier sequence;
step 1_10: calculating the distortion cost value when each carrier in the first type carrier sequence, the second type carrier sequence and the third type carrier sequence is simulated and embedded into +1, 0 or-1, for any carrier, setting the carrier as the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the mth frame in the H.265/HEVC video, and recording the distortion cost value of the carrier as rho m,n (i,j)，
Figure BDA0003517831070000041
Wherein, M is 1.ltoreq.m.ltoreq.m, M represents the total frame number of frames contained in the h.265/HEVC video, N is 1.ltoreq.n, N represents the coding tree unit contained in the M-th frame in the h.265/HEVC videoThe total number (i, j) represents the coordinate position in the coding tree unit, i is more than or equal to 1 and less than or equal to 64, j is more than or equal to 1 and less than or equal to 64, and +.>
Figure BDA0003517831070000042
Intra-block distortion cost value representing QDST coefficient of pixel point with coordinates position (i, j) in nth coding tree unit in mth frame in h.265/HEVC video, alpha represents inter-frame distortion transfer rate, GOPSize represents length of one group of pictures GOP in h.265/HEVC video, value of GOPSize is 4 or 16 or 32, q step Quantization step size representing H.265/HEVC video, cp m,n (i, j) represents texture complexity τ of a coding unit of size 4×4 where QDST coefficient of pixel point of coordinates position (i, j) in nth coding tree unit in mth frame in h.265/HEVC video is located m,n (i, j) represents a coefficient set cost value, k, of modifying QDST coefficients of pixel points having coordinate positions (i, j) in an nth coding tree unit in an mth frame in h.265/HEVC video 1 、k 2 、k 3 Are all proportional coefficients;
step 1_11: three different binary secret information sequences are randomly generated by utilizing the same embedded load rate payload and are correspondingly marked as X 1 、X 2 、X 3 The method comprises the steps of carrying out a first treatment on the surface of the Then simulating the distortion cost value when embedding +1 or 0 or-1 according to all vectors in the first class of vector sequences, and utilizing STC tool kit to store X 1 Embedding the first type carrier sequence into all carriers in the first type carrier sequence to obtain a first type carrier sequence containing the secret; similarly, X is modeled using STC toolkit based on the distorted cost value of all vectors in the second class of vector sequences when embedded with +1 or 0 or-1 2 Embedding the first carrier sequence into all carriers in the first carrier sequence to obtain a first carrier sequence containing the first secret; x is calculated by using STC toolkit according to distortion cost value of all vectors in the third class of vector sequences when being simulated and embedded into +1, 0 or-1 3 Embedding the third carrier sequence into all carriers in the third carrier sequence to obtain a third carrier sequence containing the secret; wherein, payload E (0, 0.5)]，X 1 Length of (1) is Num 1 ×payload，X 2 Length of (1) is Num 2 ×payload，X 3 Length of (1) is Num 3 ×payload，Num 1 Representing a first class of vector sequencesThe total number of vectors in (1), num 2 Represents the total number of vectors in the second class of vector sequences, num 3 Representing the total number of vectors in the third class of vector sequences;
step 1_12: according to the first class of dense carrier sequence, the first class of carrier sequence, the second class of dense carrier sequence and the second class of carrier sequence, the distortion compensation optimization is carried out, and the specific process is as follows:
step 1_12a: calculating the difference value between each dense carrier in the first type of dense carrier sequences and the corresponding carrier in the first type of carrier sequences to obtain a first difference value sequence; likewise, calculating the difference between each dense carrier in the second class of dense carrier sequences and the corresponding carrier in the second class of dense carrier sequences to obtain a second difference sequence; the difference value in the first difference value sequence is obtained by subtracting the corresponding carrier in the first type carrier sequence from the dense carrier in the first type carrier sequence, the difference value in the second difference value sequence is obtained by subtracting the corresponding carrier in the second type carrier sequence from the dense carrier in the second type carrier sequence, and each difference value in the first difference value sequence and the second difference value sequence is 0 or +1 or-1;
Step 1_12b: for any difference value in the first difference value sequence and the second difference value sequence, if the difference value is +1, when the difference value belongs to the first difference value sequence, subtracting 1 from the value of the QDST coefficient of the pixel point with index position pos+2 in the H.265/HEVC video, and adding 1 to the value of the QDST coefficient of the pixel point with index position pos+3 to realize the distortion compensation optimization of the two QDST coefficients; when the difference value belongs to a second difference value sequence, subtracting 1 from the value of the QDST coefficient of the pixel point with index position pos+8 in the H.265/HEVC video, and adding 1 to the value of the QDST coefficient of the pixel point with index position pos+12 to realize the distortion compensation optimization of the two QDST coefficients;
if the difference value is-1, when the difference value belongs to a first difference value sequence, adding 1 to the value of the QDST coefficient of the pixel point with index position pos+2 in the H.265/HEVC video and subtracting 1 to the value of the QDST coefficient of the pixel point with index position pos+3 to realize the distortion compensation optimization of the two QDST coefficients; when the difference value belongs to a second difference value sequence, adding 1 to the value of the QDST coefficient of the pixel point with index position pos+8 in the H.265/HEVC video and subtracting 1 to the value of the QDST coefficient of the pixel point with index position pos+12 to realize the distortion compensation optimization of the two QDST coefficients;
If the difference value is 0, the QDST coefficient of any pixel point in the H.265/HEVC video is not subjected to distortion compensation optimization no matter whether the difference value belongs to the first difference value sequence or the second difference value sequence;
wherein pos represents the index position of the pixel point corresponding to the difference value in the H.265/HEVC video;
step 1_13: adopting an H.265/HEVC standard encoder to perform compression encoding on an original video, replacing the original QDST coefficient at a corresponding position with the QDST coefficient after distortion compensation optimization in the transformation and quantization process of compression encoding, and performing compression encoding to obtain a code stream containing a secret video;
the specific process of extracting the secret information is as follows:
step 2_1: decoding the code stream of the video containing the secret by adopting an H.265/HEVC standard decoder to obtain a decoded video, and storing a prediction mode and depth of each coding unit in each coding tree unit in each frame in the decoded video, a PU partition mode corresponding to each coding unit and a QDST coefficient of each pixel point in each coding unit in the decoding process;
step 2_2: according to the process from the step 1_2 to the step 1_9, a first carrier sequence, a second carrier sequence and a second carrier sequence corresponding to the decoded video are obtained in the same mode;
Step 2_3: decoding the first type carrier sequence obtained in the step 2_2 by using an STC tool kit, and extracting to obtain first secret information; similarly, decoding the second class carrier sequence obtained in the step 2_2 by using an STC tool kit, and extracting to obtain second secret information; and (3) decoding the third class carrier sequence obtained in the step (2) by using the STC tool kit, and extracting to obtain third secret information.
In the step 1_6, th takes a value of 1.
In the step 1_10, alpha takes a value of 0.6 and k 1 Take the value of 4, k 2 Take the value of 1, k 3 The value is 1.
Step 1In the case of a number of a' 10,
Figure BDA0003517831070000061
is equal to the sum of absolute values of all elements in a distortion matrix E of a coding unit of size 4 x 4 where a QDST coefficient of a pixel point of coordinates (i, j) in an nth coding tree unit in an mth frame in a modified h.265/HEVC video is caused by the QDST coefficient T ·Δ·Q step ·2 6-shift H, wherein symbol "·" is a dot product operation symbol, H represents a DST transformation matrix in HEVC coding standard, H T For H transpose, Δ represents the modification matrix at the time of carrier analog embedding, the dimension of Δ is 4×4, shift represents the scaling factor of DST transform process in HEVC coding standard, +.>
Figure BDA0003517831070000071
floor () is a rounding down function, QP represents the quantization parameter of h.265/HEVC video, it_shift represents the scaling factor of the DST inverse transform process in the HEVC coding standard.
In the step 1_10, cp m,n (i,j)＝<P,|QDST m,n |>Wherein P represents a matrix obtained from statistical randomness analysis of QDST coefficients of all pixel points in a coding unit of size 4 x 4,
Figure BDA0003517831070000072
QDST m,n coefficient matrix composed of QDST coefficients representing all pixel points in coding units of size 4×4 where QDST coefficients of pixel points of coordinate position (i, j) in nth coding tree unit in mth frame in H.265/HEVC video are located, QDST m,n Is 4 x 4 in dimension, symbol'< >The symbol is an inner product operation symbol, and the symbol is an absolute value symbol.
In the step 1_10 described above,
Figure BDA0003517831070000073
wherein INF represents a positive infinite value, SCG represents QDST coefficients of pixel points with coordinates of (i, j) in an nth coding tree unit in an mth frame in H.265/HEVC videoThe sum of absolute values of all QDST coefficients in the corresponding coefficient group is a smaller constant for preventing DNNZC from being zero, flag represents a difference between the number of non-zero QDST coefficients in a 4×4 coding unit of a size where QDST coefficients of pixel points with coordinates of (i, j) are located in an nth coding tree unit in an mth.265/HEVC video before and after carrier analog embedding, flag=1 if the sum of absolute values of all QDST coefficients in the corresponding coefficient group is changed, flag=0 if the sum of absolute values is not changed, DNNZC represents a difference between the number of non-zero QDST coefficients in a 4×4 coding unit of a size where QDST coefficients of pixel points with coordinates of (i, j) are located in an nth coding tree unit in an mth.265/HEVC video before and after carrier analog embedding, and index positions of the corresponding pixel points in an h.265/HEVC video are pos is set. The carrier belongs to a first carrier sequence, and when the carrier belongs to a first carrier sequence, the coefficient group corresponding to the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the m-th frame in the H.265/HEVC video is (QDST) pos ,qdst pos+2 ,qdst pos+3 ) The carrier belongs to the second carrier sequence, and the coefficient set corresponding to the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the mth frame in the H.265/HEVC video is (QDST) pos ,qdst pos+8 ,qdst pos+12 ) When the carrier belongs to the third carrier sequence, the coefficient set corresponding to the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the mth frame in the H.265/HEVC video is (QDST) pos )，qdst pos QDST coefficient, QDST representing pixel point with coordinates position (i, j) in nth coding tree unit in mth frame in H.265/HEVC video pos+2 QDST coefficient, QDST representing pixel point with index position pos+2 in H.265/HEVC video pos+3 QDST coefficient, QDST representing pixel point with index position pos+3 in H.265/HEVC video pos+8 QDST coefficient, QDST representing pixel point with index position pos+8 in H.265/HEVC video pos+12 QDST coefficients representing pixels with index positions pos+12 in h.265/HEVC video.
Compared with the prior art, the invention has the advantages that:
1) The method designs the self-adaptive distortion cost calculation function through analyzing factors such as intra-block distortion, inter-frame distortion influence, block texture complexity, coefficient group cost and the like, and uses STC algorithm to carry out self-adaptive data hiding on QDST coefficients, so that the quality of the video picture after steganography can be effectively improved.
2) The method comprehensively considers the entropy coding process of video coding during modification, and reduces the influence of the modified QDST coefficient on the entropy coding by analyzing the factors such as the position change of the last nonzero coefficient, the sum of absolute values of coefficients, the nonzero coefficient change and the like of the coefficients in the block before and after modification, thereby reducing the increase of video code rate and improving the security of steganography.
3) The method can ensure that the secret information is effectively extracted, and the steganography process and the extraction process are combined with the HEVC video compression process, so that the method can be applied to real-time communication in real life.
Drawings
FIG. 1 is a block diagram of a general implementation of the method of the present invention;
FIG. 2a is a schematic diagram illustrating the determination of candidate blocks of a first type;
FIG. 2b is a schematic illustration of the determination of a second type of candidate block;
FIG. 2c is a schematic illustration of the determination of a third class of candidate blocks;
FIG. 3 is a diagram showing the result of the BIR of the encrypted video obtained by performing data hiding on the original video by using the DCO method;
fig. 4 is a result of a steganographic analysis detection error rate of a dense video obtained by performing data hiding on an original video using a DCO method and performing data hiding on the original video using the method of the present invention.
Detailed Description
The invention is described in further detail below with reference to the embodiments of the drawings.
The invention provides an H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization, the general implementation block diagram of which is shown in figure 1 and comprises two parts of secret information embedding and secret information extracting.
The specific process of the secret information embedding is as follows:
step 1_1: performing compression Coding on an original video by adopting an H.265/HEVC standard coder to obtain an H.265/HEVC video, and storing a prediction mode and depth of each Coding Unit (CU, coding Unit) in each frame of the H.265/HEVC video, a PU (Prediction Unit) division mode corresponding to each Coding Unit, a QDST (Quantized Discrete Sine Transform) coefficient of each pixel point in each Coding Unit and a quantized discrete sine transformation coefficient in the compression Coding process; 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 or 4×4.
Step 1_2: all frames in the h.265/HEVC video are traversed in sequence, defining the currently traversed frame as the current frame.
Step 1_3: judging whether the current frame is an I frame or not, and if the current frame is the I frame, executing the step 1_4; if the current frame is not an I-frame, step 1_8 is directly performed.
Step 1_4: all 4×4 coding units in the current frame are traversed in sequence, and the currently traversed coding unit with the size of 4×4 is defined as the current unit.
Step 1_5: classifying the current unit as a first type candidate block, a second type candidate block or a third type candidate block, or not processing the current unit, wherein the specific process is as follows:
as shown in fig. 2a, if the current unit has only the adjacent lower left coding unit of size 4×4 and the number of the prediction mode of the lower left coding unit is identified in the interval [2,26], and the adjacent directly lower 4 coding unit of size and the number of the prediction mode of the directly lower coding unit is identified in the interval [2,10], the current unit is classified as a first type candidate block, and then step 1_6 is performed.
As shown in fig. 2b, if the current unit has only the adjacent upper right coding unit of size 4×4 and the number of the prediction mode of the upper right coding unit is identified in the interval [10,34] or 1, and the adjacent right coding unit of size 4×4 and the number of the prediction mode of the right coding unit is identified in the interval [26,34], the current unit is classified as the second type candidate block, and then step 1_6 is performed.
As shown in fig. 2c, if the current unit has an adjacent 4×4 lower left coding unit and the number of prediction modes of the lower left coding unit is identified in the interval [2,26], an adjacent 4×4 lower right coding unit and the number of prediction modes of the lower right coding unit is identified in the interval [2,10], an adjacent 4×4 upper right coding unit and the number of prediction modes of the upper right coding unit is identified in the interval [10,34] or as 1, an adjacent 4×4 right coding unit and the number of prediction modes of the right coding unit is identified in the interval [26,34], an adjacent 4×4 lower right coding unit and the number of prediction modes of the right coding unit is identified in the interval [26,34], or as 0 or as 1, then the step 1_6 is performed.
For the rest of the cases, no processing is done for the current cell, and then step 1_7 is directly performed.
The above, the left lower coding unit is a coding unit with a size of 4×4 located at the left lower side of the current unit, the right lower coding unit is a coding unit with a size of 4×4 located at the right lower side of the current unit, the right upper coding unit is a coding unit with a size of 4×4 located at the right upper side of the current unit, the right lower coding unit is a coding unit with a size of 4×4 located at the right lower side of the current unit, the left lower coding unit is located at the right lower side of the current unit, the right lower coding unit is located at the right lower side of the right lower coding unit, the right upper coding unit is located at the right upper side of the right lower coding unit, and the right lower coding unit is located at the right lower side of the right lower coding unit.
Step 1_6: selecting carriers from QDST coefficients of all pixel points in a current unit, selecting QDST coefficients with absolute values larger than or equal to a set threshold th from 4 QDST coefficients of a 1 st column pixel point of the current unit (namely 4 pixel points with coordinates of (1, 1), (2, 1), (3, 1), (4, 1) in the current unit) as carriers if the current unit is a first type candidate block, classifying the carriers as first type carriers, and then executing step 1_7; if the current unit is a candidate block of the second type, selecting QDST coefficients with absolute values greater than or equal to a set threshold th from 4 QDST coefficients of 1 st row pixel points of the current unit (namely, 4 pixel points with coordinates of (1, 1), (1, 2), (1, 3), (1, 4) in the current unit) as carriers, classifying the carriers as carriers of the second type, and then executing step 1_7; if the current unit is a third type candidate block, selecting QDST coefficients with absolute values larger than or equal to a set threshold th from 16 QDST coefficients of all pixel points of the current unit as carriers, classifying the carriers into a third type carrier, and then executing step 1_7; wherein th e (0, 10), the carrier is less if the value of th is too large, the embedding capacity is less, the distortion is too large if the value of th is 0, the selection is not recommended, the embedding capacity and the distortion are considered in a compromise, and the value of th is selected to be 1 in the embodiment.
Step 1_7: and traversing the next 4×4 coding unit in the current frame as the current unit, returning to the step 1_5, continuing to execute until all the 4×4 coding units in the current frame are traversed, and then executing the step 1_8.
Step 1_8: and traversing the next frame in the H.265/HEVC video as the current frame, and returning to the step 1_3 to continue execution until all frames in the H.265/HEVC video are traversed completely.
Step 1_9: arranging all carriers classified as the first type of carriers, namely QDST coefficients, according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a first type of carrier sequence; likewise, all carriers classified as the second type of carriers, namely QDST coefficients, are arranged according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a second type of carrier sequence; and arranging all carriers classified as the third type of carriers, namely QDST coefficients, according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a third type of carrier sequence.
Step 1_10: calculating the distortion cost value when each carrier in the first type carrier sequence, the second type carrier sequence and the third type carrier sequence is simulated and embedded into +1, 0 or-1, for any carrier, setting the carrier as the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the mth frame in the H.265/HEVC video, and recording the distortion cost value of the carrier as rho m,n (i,j)，
Figure BDA0003517831070000111
Wherein, M is equal to or greater than 1 and is equal to or less than M, M represents the total frame number of frames contained in the H.265/HEVC video, N is equal to or greater than 1 and is equal to or less than N, N represents the total number of coding tree units contained in the M-th frame in the H.265/HEVC video, (i, j) represents the coordinate position in the coding tree units, i is equal to or greater than 1 and is equal to or less than 64, j is equal to or less than 1 and is equal to or less than 64, and N is equal to or less than 1>
Figure BDA0003517831070000121
An intra-block distortion cost value representing a QDST coefficient modifying a pixel point of (i, j) in a coordinate position in an nth coding tree unit in an mth frame in h.265/HEVC video, α represents an inter-frame distortion transfer rate, α takes a value of 0.6 in this embodiment, GOPSize represents a length of one group of pictures GOP in h.265/HEVC video, GOPSize takes a value of 4 or 16 or 32 in this embodiment, and GOPSize takes a value of 4, q step Quantization step size representing H.265/HEVC video, cp m,n (i, j) represents texture complexity τ of a coding unit of size 4×4 where QDST coefficient of pixel point of coordinates position (i, j) in nth coding tree unit in mth frame in h.265/HEVC video is located m,n (i, j) represents a coefficient set cost value, k, of modifying QDST coefficients of pixel points having coordinate positions (i, j) in an nth coding tree unit in an mth frame in h.265/HEVC video 1 、k 2 、k 3 Are all proportional coefficients, k is taken in this embodiment 1 ＝4、k 2 ＝1、k 3 =1, alpha is found to be 0.6 and k is found by experiment 1 ＝4、k 2 ＝1、k 3 Video embedding effect is better when=1.
Step 1_11: random generation of three different binary entries using the same embedded load rate payloadSecret information sequence is prepared and is correspondingly marked as X 1 、X 2 、X 3 The method comprises the steps of carrying out a first treatment on the surface of the Then simulating the distortion cost value when embedding +1 or 0 or-1 according to all vectors in the first class of vector sequences, and utilizing STC (Syndrome-Trellis Codes) tool package to store X 1 Embedding the first type carrier sequence into all carriers in the first type carrier sequence to obtain a first type carrier sequence containing the secret; similarly, X is modeled using STC toolkit based on the distorted cost value of all vectors in the second class of vector sequences when embedded with +1 or 0 or-1 2 Embedding the first carrier sequence into all carriers in the first carrier sequence to obtain a first carrier sequence containing the first secret; x is calculated by using STC toolkit according to distortion cost value of all vectors in the third class of vector sequences when being simulated and embedded into +1, 0 or-1 3 Embedding the third carrier sequence into all carriers in the third carrier sequence to obtain a third carrier sequence containing the secret; wherein, payload E (0, 0.5)]In this embodiment, payload takes 0.4, X 1 Length of (1) is Num 1 ×payload，X 2 Length of (1) is Num 2 ×payload，X 3 Length of (1) is Num 3 ×payload，Num 1 Represents the total number of vectors in the first type of vector sequence, num 2 Represents the total number of vectors in the second class of vector sequences, num 3 Representing the total number of vectors in the third class of vector sequences.
Step 1_12: according to the first class of dense carrier sequence, the first class of carrier sequence, the second class of dense carrier sequence and the second class of carrier sequence, the distortion compensation optimization is carried out, and the specific process is as follows:
step 1_12a: calculating the difference value between each dense carrier in the first type of dense carrier sequences and the corresponding carrier in the first type of carrier sequences to obtain a first difference value sequence; likewise, calculating the difference between each dense carrier in the second class of dense carrier sequences and the corresponding carrier in the second class of dense carrier sequences to obtain a second difference sequence; the difference value in the first difference sequence is obtained by subtracting the corresponding carrier in the first type carrier sequence from the dense carrier in the first type carrier sequence, the difference value in the second difference sequence is obtained by subtracting the corresponding carrier in the second type carrier sequence from the dense carrier in the second type carrier sequence, and each difference value in the first difference sequence and the second difference sequence is 0 or +1 or-1.
Step 1_12b: for any difference value in the first difference value sequence and the second difference value sequence, if the difference value is +1, when the difference value belongs to the first difference value sequence, subtracting 1 from the value of the QDST coefficient of the pixel point with index position pos+2 in the H.265/HEVC video, and adding 1 to the value of the QDST coefficient of the pixel point with index position pos+3 to realize the distortion compensation optimization of the two QDST coefficients; when the difference value belongs to the second difference value sequence, the value of the QDST coefficient of the pixel point with index position pos+8 in the H.265/HEVC video is reduced by 1, and the value of the QDST coefficient of the pixel point with index position pos+12 is increased by 1, so that the distortion compensation optimization of the two QDST coefficients is realized.
If the difference value is-1, when the difference value belongs to a first difference value sequence, adding 1 to the value of the QDST coefficient of the pixel point with index position pos+2 in the H.265/HEVC video and subtracting 1 to the value of the QDST coefficient of the pixel point with index position pos+3 to realize the distortion compensation optimization of the two QDST coefficients; when the difference value belongs to the second difference value sequence, the value of the QDST coefficient of the pixel point with index position pos+8 in the H.265/HEVC video is added with 1, and the value of the QDST coefficient of the pixel point with index position pos+12 is subtracted with 1, so that the distortion compensation optimization of the two QDST coefficients is realized.
If the difference value is 0, no distortion compensation optimization is performed on the QDST coefficient of any pixel point in the h.265/HEVC video, no matter whether the difference value belongs to the first difference value sequence or the second difference value sequence.
Wherein pos represents the index position of the pixel point corresponding to the difference value in the h.265/HEVC video.
For example: QDST coefficients of pixels with index positions of pos in the H.265/HEVC video, QDST coefficients of pixels with index positions of pos+2, QDST coefficients of pixels with index positions of pos+3 correspond to 5,6 and 7 respectively, QDST coefficients of pixels with index positions of pos in the H.265/HEVC video belong to a first carrier sequence, QDST coefficients obtained after embedding secret information, namely a dense carrier, of QDST coefficients obtained after embedding secret information, and the difference value obtained is +1, so that the QDST coefficients of pixels with index positions of pos+2 in the H.265/HEVC video and QDST coefficients of pixels with index positions of pos+3 correspond to 5 and 8 after distortion compensation optimization. I.e. the set of coefficients at pre-steganography pos is (5, 6, 7), and the set of coefficients at post steganography and distortion compensation optimization pos is (6,5,8).
Step 1_13: and (3) carrying out compression coding on the original video by adopting an H.265/HEVC standard coder, replacing the original QDST coefficient at the corresponding position by the QDST coefficient after distortion compensation optimization in the transformation and quantization process of the compression coding, and carrying out compression coding to obtain a code stream containing the secret video.
The specific process of extracting the secret information is as follows:
step 2_1: and decoding the coded stream of the video containing the secret by adopting an H.265/HEVC standard decoder to obtain a decoded video, and storing a prediction mode and depth of each coding unit in each coding tree unit in each frame in the decoded video, a PU partition mode corresponding to each coding unit and a QDST coefficient of each pixel point in each coding unit in the decoding process.
Step 2_2: according to the process from step 1_2 to step 1_9, the first type of carrier sequence, the second type of carrier sequence and the second type of carrier sequence corresponding to the decoded video are obtained in the same manner.
Step 2_3: decoding the first type carrier sequence obtained in the step 2_2 by using an STC tool kit, and extracting to obtain first secret information; similarly, decoding the second class carrier sequence obtained in the step 2_2 by using an STC tool kit, and extracting to obtain second secret information; and (3) decoding the third class carrier sequence obtained in the step (2) by using the STC tool kit, and extracting to obtain third secret information.
In this embodiment, in step 1_10,
Figure BDA0003517831070000141
is equal to the sum of absolute values of all elements in a distortion matrix E of a coding unit of size 4 x 4 where a QDST coefficient of a pixel point of coordinates (i, j) in an nth coding tree unit in an mth frame in a modified h.265/HEVC video is caused by the QDST coefficient T ·Δ·Q step ·2 6 -shift H, where symbol "·" is a dot-product operator, H represents the DST transform matrix in the HEVC coding standard,
Figure BDA0003517831070000142
H T the vector is transposed to H, delta represents a modification matrix when the vector is simulated and embedded, the dimension of delta is 4 multiplied by 4, delta is given in the HEVC coding process, delta is actually determined by the modification quantity when the vector is simulated and embedded in +1 or 0 or-1, if the modification quantity of the vector (when the index position of a pixel point corresponding to the vector is pos) is t, when the vector belongs to a first-class vector sequence, the compensation modification quantity of the QDST coefficient of the pixel point with the index position pos+2 and the compensation modification quantity of the QDST coefficient of the pixel point with the index position pos+3 is t>
Figure BDA0003517831070000151
When the carrier belongs to the second carrier sequence, the compensation modification quantity of the QDST coefficient of the pixel point with the index position of pos+8 and the QDST coefficient of the pixel point with the index position of pos+12 is t,
Figure BDA0003517831070000152
when the vector belongs to the third class of vector sequences, and (2) >
Figure BDA0003517831070000153
t has a value of 1 or-1 or 0, shift represents the scaling factor of the DST transform process in HEVC coding standard, +.>
Figure BDA0003517831070000154
floor () is a rounding down function, QP represents the quantization parameter of h.265/HEVC video, it_shift represents the scaling factor of the DST inverse transform process in the HEVC coding standard.
In the present embodiment, in step 1_10, cp m,n (i,j)＝<P,|QDST m,n |>Wherein P represents a matrix obtained from a statistical randomness analysis of QDST coefficients of all pixel points in a 4 x 4 size coding unit for characterizing a maximum distribution statistical feature,
Figure BDA0003517831070000155
QDST m,n coefficient matrix composed of QDST coefficients representing all pixel points in coding units of size 4×4 where QDST coefficients of pixel points of coordinate position (i, j) in nth coding tree unit in mth frame in H.265/HEVC video are located, QDST m,n Is 4 x 4 in dimension, symbol'< >The symbol is an inner product operation symbol, and the symbol is an absolute value symbol.
In this embodiment, in step 1_10,
Figure BDA0003517831070000156
wherein INF represents a positive infinitely variable value, SCG represents whether the position of the last non-zero QDST coefficient in the coding unit of size 4×4 where the QDST coefficient of the pixel point of coordinate position (i, j) in the nth coding tree unit in the h.265/HEVC video corresponds to the sum of absolute values of all QDST coefficients in the coefficient group corresponding to the QDST coefficient of the pixel point, constant is a smaller constant used to prevent DNNZC from being zero, if the value of constant is 0.001, flag represents whether the position of the corresponding coefficient of the QDST coefficient of the pixel point of coordinate position (i, j) in the nth coding tree unit in the h.265/HEVC video before and after carrier analog embedding is set to be zero in the coding unit of size 4×4 where the index of the pixel point of coordinate position (i, j) in the nth coding tree unit in the h.265/HEVC video corresponds to the index of size 4×4 where the index of the pixel point of the QDST coefficient of the pixel point in the nth coding tree unit in the h.265/HEVC video corresponds to the carrier analog embedding is set to be zero: the carrier belongs to a first carrier sequence, and when the carrier belongs to a first carrier sequence, the coefficient group corresponding to the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the m-th frame in the H.265/HEVC video is (QDST) pos ,qdst pos+2 ,qdst pos+3 ) The carrier belongs to the second carrier sequence, and the coefficient set corresponding to the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the mth frame in the H.265/HEVC video is (QDST) pos ,qdst pos+8 ,qdst pos+12 ) The vector belongs to the H.265/HEVC video when the third class of vector sequencesThe coefficient set corresponding to QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the mth frame is (QDST) pos )，qdst pos QDST coefficient, QDST representing pixel point with coordinates position (i, j) in nth coding tree unit in mth frame in H.265/HEVC video pos+2 QDST coefficient, QDST representing pixel point with index position pos+2 in H.265/HEVC video pos+3 QDST coefficient, QDST representing pixel point with index position pos+3 in H.265/HEVC video pos+8 QDST coefficient, QDST representing pixel point with index position pos+8 in H.265/HEVC video pos+12 QDST coefficients representing pixels with index positions pos+12 in h.265/HEVC video.
To further illustrate the feasibility and effectiveness of the process of the invention, experiments were performed.
Here, 4 different video sequences with a resolution of 1920×1080 are selected, namely "baseball drive", "bqterce", "catus" and "ParkScene", respectively. The method is carried out under H.265/HEVC reference software HM16.15, and uses a self-contained encoder_lowdelay_P_main.cfg configuration file for encoding, wherein the encoding structure of a video sequence is IPPP, the length of a group of pictures GOP is 4, the number of frames is 40 (i.e. the first 40 frames of each video sequence are respectively selected), the encoding quantization parameter QP is fixed to be 25, and the rest encoding parameters are consistent with the original configuration file.
To further illustrate the effectiveness of the method of the present invention, five indicators, namely Peak Signal-to-Noise Ratio (PSNR), structural similarity (Structural Similarity, SSIM), mean square error (Mean Squared Error, MSE), bit rate increase (Bit Increase Rate, BIR), and steganographic analysis detection error rate (detection error rate), are used herein to evaluate the visual quality impact of HEVC video data hiding, wherein the larger the PSNR and SSIM values are, the better the video picture quality is; the smaller the MSE, the better the video picture quality; the smaller BIR is, the smaller the influence of the steganography method on the video code rate is; the greater the steganalysis detection error rate, the more the steganalysis method can fool the steganalysis detection.
Table 1 shows a comparison of the results on PSNR, SSIM, MSE of a video obtained by compressing an original video using only an HM encoder, a dense video obtained by data hiding an original video using a DCO method, and a dense video obtained by data hiding an original video using the method of the present invention.
Table 1 comparison of PSNR, SSIM, MSE obtained by three methods
Figure BDA0003517831070000171
In Table 1, "Original" refers to a method of compressing an Original video using an HM encoder, and the DCO method refers to a method of H.265/HEVC data hiding without intra-frame distortion drift by Yongxia L, shuyang L, hongguo Z, et al A new data hiding method for H.265/HEVC video streams without intra-frame distortion drift [ J ]. Multi-media Tools & Applications,2018.
As can be seen from the data listed in table 1, the video sequence containing the secret modified by the method of the present invention is superior to the comparison method in video picture quality.
Fig. 3 shows the result of the data hiding of the original video using the DCO method and the result of the data hiding of the original video using the method of the present invention on the BIR, and fig. 4 shows the result of the data hiding of the original video using the DCO method and the result of the data hiding of the original video using the method of the present invention on the steganography analysis detection error rate.
As can be seen from the data shown in fig. 3, the code rate of the video sequence containing the secret, which is modified by the method of the present invention, has a smaller influence than that of the comparison method. As can be seen from the data listed in FIG. 4, the steganalysis detection error rate of most of the video sequences containing the secret information modified by the method of the present invention is higher than that of the comparison method, which indicates that the security of the method of the present invention is higher.

Claims (6)

1. An H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization comprises two parts of secret information embedding and secret information extracting; the method is characterized in that:
the specific process of the secret information embedding is as follows:
Step 1_1: performing compression coding on an original video by adopting an H.265/HEVC standard coder to obtain an H.265/HEVC video, and storing a prediction mode and depth of each coding unit in each coding tree unit in each frame in the H.265/HEVC video, a PU partition mode corresponding to each coding unit and a QDST coefficient of each pixel point in each coding unit in the compression coding process; 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 or 4×4;
step 1_2: traversing all frames in the H.265/HEVC video in sequence, and defining the currently traversed frame as a current frame;
step 1_3: judging whether the current frame is an I frame or not, and if the current frame is the I frame, executing the step 1_4; if the current frame is not the I frame, directly executing the step 1_8;
step 1_4: traversing all coding units with the size of 4 multiplied by 4 in the current frame in sequence, and defining the currently traversed coding units with the size of 4 multiplied by 4 as current units;
step 1_5: classifying the current unit as a first type candidate block, a second type candidate block or a third type candidate block, or not processing the current unit, wherein the specific process is as follows:
if the current unit has only adjacent 4×4 lower left coding units and the number of prediction modes of the lower left coding units is identified in the interval [2,26], and adjacent 4×4 lower right coding units and the number of prediction modes of the lower right coding units is identified in the interval [2,10], classifying the current unit as a first type candidate block, and then performing step 1_6;
If the current unit has only the adjacent upper right coding unit of size 4×4 and the number of the prediction mode of the upper right coding unit is identified in the interval [10,34] or 1, and the adjacent right coding unit of size 4×4 and the number of the prediction mode of the right coding unit is identified in the interval [26,34], classifying the current unit as a second type candidate block, and then performing step 1_6;
if the current unit has adjacent 4×4 lower left coding unit and the number of prediction modes of the lower left coding unit is within the interval [2,26], and adjacent 4×4 lower right coding unit and the number of prediction modes of the lower right coding unit is within the interval [2,10], and adjacent 4×4 upper right coding unit and the number of prediction modes of the upper right coding unit is within the interval [10,34] or 1, and adjacent 4×4 right coding unit and the number of prediction modes of the lower right coding unit is within the interval [26,34], and adjacent 4×4 lower right coding unit and the number of prediction modes of the lower right coding unit is within the interval [26,34] or 0 or 1, classifying the current unit as a third type candidate block, and then performing step 1_6;
For the rest of the cases, the current unit is not processed, and then step 1_7 is directly performed;
the above, the lower left coding unit is a coding unit with a size of 4×4 located right below the current unit, the lower right coding unit is a coding unit with a size of 4×4 located right below the current unit, the upper right coding unit is a coding unit with a size of 4×4 located right above the current unit, the lower right coding unit is a coding unit with a size of 4×4 located right above the current unit, and the lower right coding unit is a coding unit with a size of 4×4 located right below the current unit;
step 1_6: selecting carriers from the QDST coefficients of all pixel points in the current unit, if the current unit is a first type candidate block, selecting QDST coefficients with absolute values larger than or equal to a set threshold th from 4 QDST coefficients of the 1 st column pixel point of the current unit as carriers, classifying the carriers as first type carriers, and then executing step 1_7; if the current unit is a second type candidate block, selecting QDST coefficients with absolute values larger than or equal to a set threshold th from 4 QDST coefficients of a 1 st row pixel point of the current unit as carriers, classifying the carriers as a second type carrier, and then executing step 1_7; if the current unit is a third type candidate block, selecting QDST coefficients with absolute values larger than or equal to a set threshold th from 16 QDST coefficients of all pixel points of the current unit as carriers, classifying the carriers into a third type carrier, and then executing step 1_7; wherein th E (0, 10);
Step 1_7: traversing the next 4×4 coding unit in the current frame as the current unit, returning to step 1_5, continuing to execute until all the 4×4 coding units in the current frame are traversed, and then executing step 1_8;
step 1_8: traversing the next frame in the H.265/HEVC video as the current frame, and returning to the step 1_3 to continue execution until all frames in the H.265/HEVC video are traversed completely;
step 1_9: arranging all carriers classified as the first type of carriers, namely QDST coefficients, according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a first type of carrier sequence; likewise, all carriers classified as the second type of carriers, namely QDST coefficients, are arranged according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a second type of carrier sequence; arranging all carriers classified as third-class carriers, namely QDST coefficients, according to the sequence of index positions of corresponding pixel points in the H.265/HEVC video to form a third-class carrier sequence;
step 1_10: calculating the distortion cost value when each carrier in the first type carrier sequence, the second type carrier sequence and the third type carrier sequence is simulated and embedded into +1, 0 or-1, for any carrier, setting the carrier as the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the mth frame in the H.265/HEVC video, and recording the distortion cost value of the carrier as rho m,n (i,j)，
Figure FDA0003517831060000031
Wherein, M is equal to or greater than 1 and is equal to or less than M, M represents the total frame number of frames contained in the H.265/HEVC video, N is equal to or greater than 1 and is equal to or less than N, N represents the total number of coding tree units contained in the M-th frame in the H.265/HEVC video, (i, j) represents the coordinate position in the coding tree units, i is equal to or greater than 1 and is equal to or less than 64, j is equal to or less than 1 and is equal to or less than 64, and N is equal to or less than 1>
Figure FDA0003517831060000032
Intra-block distortion cost value representing QDST coefficient of pixel point with coordinates position (i, j) in nth coding tree unit in mth frame in h.265/HEVC video, alpha represents inter-frame distortion transfer rate, GOPSize represents length of one group of pictures GOP in h.265/HEVC video, value of GOPSize is 4 or 16 or 32, q step Quantization step size representing H.265/HEVC video, cp m,n (i, j) represents texture complexity τ of a coding unit of size 4×4 where QDST coefficient of pixel point of coordinates position (i, j) in nth coding tree unit in mth frame in h.265/HEVC video is located m,n (i, j) represents a coefficient set cost value, k, of modifying QDST coefficients of pixel points having coordinate positions (i, j) in an nth coding tree unit in an mth frame in h.265/HEVC video 1 、k 2 、k 3 Are all proportional coefficients;
step 1_11: three different binary secret information sequences are randomly generated by utilizing the same embedded load rate payload and are correspondingly marked as X 1 、X 2 、X 3 The method comprises the steps of carrying out a first treatment on the surface of the Then simulating the distortion cost value when embedding +1 or 0 or-1 according to all vectors in the first class of vector sequences, and utilizing STC tool kit to store X 1 Embedding the first type carrier sequence into all carriers in the first type carrier sequence to obtain a first type carrier sequence containing the secret; similarly, X is modeled using STC toolkit based on the distorted cost value of all vectors in the second class of vector sequences when embedded with +1 or 0 or-1 2 Embedding the first carrier sequence into all carriers in the first carrier sequence to obtain a first carrier sequence containing the first secret; x is calculated by using STC toolkit according to distortion cost value of all vectors in the third class of vector sequences when being simulated and embedded into +1, 0 or-1 3 Embedding the third carrier sequence into all carriers in the third carrier sequence to obtain a third carrier sequence containing the secret; wherein, payload E (0, 0.5)]，X 1 Length of (1) is Num 1 ×payload，X 2 Length of (1) is Num 2 ×payload，X 3 Length of (1) is Num 3 ×payload，Num 1 Represents the total number of vectors in the first type of vector sequence, num 2 Representing a second class of vector sequencesTotal number of vectors, num 3 Representing the total number of vectors in the third class of vector sequences;
step 1_12: according to the first class of dense carrier sequence, the first class of carrier sequence, the second class of dense carrier sequence and the second class of carrier sequence, the distortion compensation optimization is carried out, and the specific process is as follows:
step 1_12a: calculating the difference value between each dense carrier in the first type of dense carrier sequences and the corresponding carrier in the first type of carrier sequences to obtain a first difference value sequence; likewise, calculating the difference between each dense carrier in the second class of dense carrier sequences and the corresponding carrier in the second class of dense carrier sequences to obtain a second difference sequence; the difference value in the first difference value sequence is obtained by subtracting the corresponding carrier in the first type carrier sequence from the dense carrier in the first type carrier sequence, the difference value in the second difference value sequence is obtained by subtracting the corresponding carrier in the second type carrier sequence from the dense carrier in the second type carrier sequence, and each difference value in the first difference value sequence and the second difference value sequence is 0 or +1 or-1;
Step 1_12b: for any difference value in the first difference value sequence and the second difference value sequence, if the difference value is +1, when the difference value belongs to the first difference value sequence, subtracting 1 from the value of the QDST coefficient of the pixel point with index position pos+2 in the H.265/HEVC video, and adding 1 to the value of the QDST coefficient of the pixel point with index position pos+3 to realize the distortion compensation optimization of the two QDST coefficients; when the difference value belongs to a second difference value sequence, subtracting 1 from the value of the QDST coefficient of the pixel point with index position pos+8 in the H.265/HEVC video, and adding 1 to the value of the QDST coefficient of the pixel point with index position pos+12 to realize the distortion compensation optimization of the two QDST coefficients;
if the difference value is-1, when the difference value belongs to a first difference value sequence, adding 1 to the value of the QDST coefficient of the pixel point with index position pos+2 in the H.265/HEVC video and subtracting 1 to the value of the QDST coefficient of the pixel point with index position pos+3 to realize the distortion compensation optimization of the two QDST coefficients; when the difference value belongs to a second difference value sequence, adding 1 to the value of the QDST coefficient of the pixel point with index position pos+8 in the H.265/HEVC video and subtracting 1 to the value of the QDST coefficient of the pixel point with index position pos+12 to realize the distortion compensation optimization of the two QDST coefficients;
If the difference value is 0, the QDST coefficient of any pixel point in the H.265/HEVC video is not subjected to distortion compensation optimization no matter whether the difference value belongs to the first difference value sequence or the second difference value sequence;
wherein pos represents the index position of the pixel point corresponding to the difference value in the H.265/HEVC video;
step 1_13: adopting an H.265/HEVC standard encoder to perform compression encoding on an original video, replacing the original QDST coefficient at a corresponding position with the QDST coefficient after distortion compensation optimization in the transformation and quantization process of compression encoding, and performing compression encoding to obtain a code stream containing a secret video;
the specific process of extracting the secret information is as follows:
step 2_1: decoding the code stream of the video containing the secret by adopting an H.265/HEVC standard decoder to obtain a decoded video, and storing a prediction mode and depth of each coding unit in each coding tree unit in each frame in the decoded video, a PU partition mode corresponding to each coding unit and a QDST coefficient of each pixel point in each coding unit in the decoding process;
step 2_2: according to the process from the step 1_2 to the step 1_9, a first carrier sequence, a second carrier sequence and a second carrier sequence corresponding to the decoded video are obtained in the same mode;
Step 2_3: decoding the first type carrier sequence obtained in the step 2_2 by using an STC tool kit, and extracting to obtain first secret information; similarly, decoding the second class carrier sequence obtained in the step 2_2 by using an STC tool kit, and extracting to obtain second secret information; and (3) decoding the third class carrier sequence obtained in the step (2) by using the STC tool kit, and extracting to obtain third secret information.
2. The h.265/HEVC video adaptive information hiding method based on distortion compensation optimization according to claim 1, wherein th takes a value of 1 in the step 1_6.
3. Distortion compensation based optimization as claimed in claim 1The H.265/HEVC video self-adaptive information hiding method is characterized in that in the step 1_10, alpha takes a value of 0.6 and k 1 Take the value of 4, k 2 Take the value of 1, k 3 The value is 1.
4. The distortion compensation optimization-based h.265/HEVC video adaptive information hiding method according to any one of claims 1 to 3, characterized in that in said step 1_10,
Figure FDA0003517831060000051
is equal to the sum of absolute values of all elements in a distortion matrix E of a coding unit of size 4 x 4 where a QDST coefficient of a pixel point of coordinates (i, j) in an nth coding tree unit in an mth frame in a modified h.265/HEVC video is caused by the QDST coefficient T ·Δ·Q step ·2 6 -shift H, wherein symbol "·" is a dot product operation symbol, H represents a DST transformation matrix in HEVC coding standard, H T For H transpose, Δ represents the modification matrix at the time of carrier analog embedding, the dimension of Δ is 4×4, shift represents the scaling factor of DST transform process in HEVC coding standard, +.>
Figure FDA0003517831060000061
floor () is a rounding down function, QP represents the quantization parameter of h.265/HEVC video, it_shift represents the scaling factor of the DST inverse transform process in the HEVC coding standard.
5. The h.265/HEVC video adaptive information hiding method based on distortion compensation optimization according to claim 4, characterized in that in said step 1_10, cp m,n (i,j)＝<P,|QDST m,n |>Wherein P represents a matrix obtained from statistical randomness analysis of QDST coefficients of all pixel points in a coding unit of size 4 x 4,
Figure FDA0003517831060000062
QDST m,n representing in H.265/HEVC videoCoefficient matrix composed of QDST coefficients of all pixel points in coding unit of size 4×4 where QDST coefficients of pixel points of (i, j) coordinate position in nth coding tree unit in mth frame are located, QDST m,n Is 4 x 4 in dimension, symbol'<>The symbol is an inner product operation symbol, and the symbol is an absolute value symbol.
6. The h.265/HEVC video adaptive information hiding method based on distortion compensation optimization according to claim 5, characterized in that in said step 1_10,
Figure FDA0003517831060000063
Wherein INF represents a positive infinite value, SCG represents whether the position of the last non-zero QDST coefficient in the coding unit of size 4×4 where the QDST coefficient of the pixel point of the n-th coding tree unit of the h.265/HEVC video is located in the m-th frame of the h.265/HEVC video, if so, flag=1, if not, flag=0, DNNZC represents that the sum of absolute values of all QDST coefficients in the coefficient group corresponding to the QDST coefficient of the pixel point of the (i, j) in the m-th coding tree unit of the h.265/HEVC video is zero, flag represents that the position of the QDST coefficient of the pixel point of the (i, j) in the n-th coding tree unit of the h.265/HEVC video is located in the m-th frame of the carrier analog embedding before and after, and if so, flag represents that the position of the QDST coefficient of the pixel point of the (i, j) in the n-th coding tree unit of the h.265/HEVC video is located in the size 4×4 where the corresponding to the QDST coefficient of the pixel point of the n-th coding tree unit of the h.265/HEVC video is located in the carrier analog embedding is located in the non-zero, if so, if not, flag=0, if so, if not: the carrier belongs to a first carrier sequence, and when the carrier belongs to a first carrier sequence, the coefficient group corresponding to the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the m-th frame in the H.265/HEVC video is (QDST) pos ,qdst pos+2 ,qdst pos+3 ) The carrier belongs to the second carrier sequence, and the coefficient set corresponding to the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the mth frame in the H.265/HEVC video is (QDST) pos ,qdst pos+8 ,qdst pos+12 ) The carrier belongs to the firstThe coefficient set corresponding to the QDST coefficient of the pixel point with the coordinate position (i, j) in the nth coding tree unit in the m-th frame in the H.265/HEVC video under the three carrier sequences is (QDST) pos )，qdst pos QDST coefficient, QDST representing pixel point with coordinates position (i, j) in nth coding tree unit in mth frame in H.265/HEVC video pos+2 QDST coefficient, QDST representing pixel point with index position pos+2 in H.265/HEVC video pos+3 QDST coefficient, QDST representing pixel point with index position pos+3 in H.265/HEVC video pos+8 QDST coefficient, QDST representing pixel point with index position pos+8 in H.265/HEVC video pos+12 QDST coefficients representing pixels with index positions pos+12 in h.265/HEVC video. />
CN202210172234.9A 2022-02-24 2022-02-24 H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization Active CN114727112B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210172234.9A CN114727112B (en) 2022-02-24 2022-02-24 H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210172234.9A CN114727112B (en) 2022-02-24 2022-02-24 H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization

Publications (2)

Publication Number Publication Date
CN114727112A CN114727112A (en) 2022-07-08
CN114727112B true CN114727112B (en) 2023-05-16

Family

ID=82236578

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210172234.9A Active CN114727112B (en) 2022-02-24 2022-02-24 H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization

Country Status (1)

Country Link
CN (1) CN114727112B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105049871A (en) * 2015-07-13 2015-11-11 宁波大学 HEVC-based audio information embedding method and extraction and reconstruction method
CN106210717A (en) * 2016-08-05 2016-12-07 北京航空航天大学 A kind of HEVC decoding complex degree control method based on saliency
CN107318022A (en) * 2017-06-27 2017-11-03 郑州师范学院 A kind of video steganography method based on the undistorted drift of H.265 standard
CN109803143A (en) * 2018-10-16 2019-05-24 曙光云计算集团有限公司 A kind of pair of video carries out the hiding method of frame information and device

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8477990B2 (en) * 2010-03-05 2013-07-02 Digimarc Corporation Reducing watermark perceptibility and extending detection distortion tolerances

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105049871A (en) * 2015-07-13 2015-11-11 宁波大学 HEVC-based audio information embedding method and extraction and reconstruction method
CN106210717A (en) * 2016-08-05 2016-12-07 北京航空航天大学 A kind of HEVC decoding complex degree control method based on saliency
CN107318022A (en) * 2017-06-27 2017-11-03 郑州师范学院 A kind of video steganography method based on the undistorted drift of H.265 standard
CN109803143A (en) * 2018-10-16 2019-05-24 曙光云计算集团有限公司 A kind of pair of video carries out the hiding method of frame information and device

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
朱银芳.一种综合带失真补偿量化和扩频的水印算法. 计算机应用与软件.2013,第30卷(第8期),332-335. *

Also Published As

Publication number Publication date
CN114727112A (en) 2022-07-08

Similar Documents

Publication Publication Date Title
Lee et al. High-payload image hiding with quality recovery using tri-way pixel-value differencing
Shi et al. Lossless data hiding: Fundamentals, algorithms, and applications
CN110677694B (en) Video encryption method for resisting contour attack
CN107040787A (en) The 3D HEVC inter-frame information hidden methods that a kind of view-based access control model is perceived
CN108366259A (en) A kind of reversible steganography method of three coefficient of H.264/AVC video of high-fidelity
CN108965887B (en) Video information hiding method and device based on inter-block decoupling
CN109257521A (en) A kind of STC Information Hiding Algorithms
CN104581176A (en) H.264/AVC (advanced video coding) compressed domain robust video watermark embedding and extracting methods free from intra-frame error drift
CN109819260B (en) Video steganography method and device based on multi-embedded domain fusion
Ge et al. Oblivious video watermarking scheme with adaptive embedding mechanism
Niu et al. Multiresolution watermarking for video based on gray-level digital watermark
Tsai et al. Adaptive data hiding for vector quantization images based on overlapping codeword clustering
CN109427032B (en) Scalable video watermark embedding and extracting method based on compressed domain
CN113329229A (en) High-capacity hiding method for H.265 video information with high-efficiency fidelity
CN114727112B (en) H.265/HEVC video self-adaptive information hiding method based on distortion compensation optimization
Sakazawa et al. H. 264 native video watermarking method
Vivek et al. Video steganography using chaos encryption algorithm with high efficiency video coding for data hiding
CN108683921B (en) Video reversible information hiding method based on zero quantization DCT coefficient group
CN108810550B (en) Video watermark embedding and extracting method based on interlayer prediction
Sakib et al. A robust DWT-based compressed domain video watermarking technique
CN113507546A (en) Image encryption method based on DCT-64
Jiaji et al. An information hiding algorithm for HEVC based on intra prediction mode and block code
CN115334316A (en) HEVC video self-adaptive steganography method without intra-frame distortion drift
Venugopala et al. Evaluation of video watermarking algorithms on mobile device
Naveen et al. Image Steganography Method for Securing Multiple Images using LSB–GA

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant