CN113840146B

CN113840146B - Video information hiding method and system

Info

Publication number: CN113840146B
Application number: CN202111156556.6A
Authority: CN
Inventors: 刘熙尧; 石凯悦; 张昊; 张健; 贺建飚; 张伟
Original assignee: Hunan Junhan Information Technology Co ltd; Central South University
Current assignee: Hunan Junhan Information Technology Co ltd; Central South University
Priority date: 2021-09-30
Filing date: 2021-09-30
Publication date: 2022-10-21
Anticipated expiration: 2041-09-30
Also published as: CN113840146A

Abstract

The invention relates to a video information hiding method and a system, in the process of coding each I frame of a video, firstly, a coding tree unit is taken as a basis, the coding blocks are divided according to the structure of a quadtree and a multi-type tree, then, based on preset coding parameters and the mapping rule of hidden information, the optimal coding parameters are determined according to the value of the current secret information to be hidden, finally, when the coding block with the preset size is coded for the first time, the current secret information to be hidden is embedded into the coding block with the preset size by utilizing the optimal coding parameters, further, the quantized transformation coefficient to be transmitted is not directly modified to realize information hiding, but the secret information is hidden into the coding block by selecting the optimal coding parameters in the coding process, the coding process of the coding block is to code by adopting the modulated information, the condition that the final coefficient is inconsistent with the original coefficient can not occur, therefore, the adjacent area can not be influenced, and the video quality can be ensured.

Description

Video information hiding method and system

Technical Field

The present invention relates to the field of video information hiding technologies, and in particular, to a video information hiding method and system.

Background

With the rapid development of the internet and the popularity of short videos, security problems such as video copyright protection are increasingly highlighted. Video information hiding technology is one of the key technologies for providing various video services, such as video authentication and enhancement, and refers to embedding different types of feature information into a video stream for security authentication, such as watermarking. Early digital video information hiding algorithms were mainly based on embedding watermark information into whole pixels of video frames with little content dependency. With the development of the technology, video information hiding technology starts to focus on the combination with video content, and more information hiding is realized on the basis of analyzing the video content (such as video scenes, video objects, motion vectors and the like). From the viewpoint of video use, a video information hiding algorithm combined with a video coding standard gets more and more attention of scholars, and therefore, the video information hiding algorithm based on the coding standard is a current or later development direction.

With the update iteration of the video coding standard, the information hiding algorithm based on the video coding standard is continuously updated. However, from h.264/AVC to h.265/HEVC, the existing information hiding algorithms basically use prediction mode, transform coefficients and the size of the coding block as the embedded carrier, wherein the Transform coefficients are most classical to modify, because Discrete Cosine Transform (DCT) is always an indispensable step in the coding standard, such as JPEG, MPEG, h.264/AVC and h.265/HEVC. The method of embedding information into a Quantized Transform Coefficient (QTC) is most widely applied to an information hiding algorithm because its embedding capacity is relatively large and robustness is good, but distortion drift occurs when information hiding is performed by modifying the quantized transform coefficient, so that distortion caused by embedding information is propagated to other blocks and other frames of a current frame, and finally, the distortion is accumulated to cause a serious influence on video quality, so that video quality is deteriorated.

Accordingly, a video information hiding method and system capable of guaranteeing video quality are needed.

Disclosure of Invention

The invention aims to provide a video information hiding method and a video information hiding system, wherein in the process of coding a video, secret information is hidden through selecting an optimal coding parameter, so that the video quality can be ensured while the video information is hidden.

In order to achieve the purpose, the invention provides the following scheme:

a method of hiding video information, the hiding method comprising:

in encoding each I-frame of a video, the following steps are performed:

on the basis of the coding tree unit, dividing coding blocks according to the structure of a quadtree and a multi-type tree to obtain coding blocks with different sizes;

determining an optimal encoding parameter according to the value of the current secret information to be hidden based on a preset encoding parameter and a mapping rule of the hidden information; the current secret information to be hidden is secret information corresponding to a coding block with a preset size;

when the coding block with the preset size is coded for the first time, the current secret information to be hidden is embedded into the coding block with the preset size by utilizing the optimal coding parameters.

A video information hiding system, the hiding system comprising:

in the process of coding each I frame of the video, the following functional modules are executed:

the dividing module is used for dividing the coding blocks according to the structure of the quadtree and the multi-type tree on the basis of the coding tree unit to obtain the coding blocks with different sizes;

the determining module is used for determining the optimal coding parameter according to the value of the current secret information to be hidden based on the preset coding parameter and the mapping rule of the hidden information; the current secret information to be hidden is secret information corresponding to a coding block with a preset size;

and the embedding module is used for embedding the current secret information to be hidden into the coding block with the preset size by utilizing the optimal coding parameter when the coding block with the preset size is coded for the first time.

According to the specific embodiment provided by the invention, the invention discloses the following technical effects:

the invention is used for providing a video information hiding method and a system, and the following steps are executed in the process of coding each I frame of a video: the method comprises the steps of dividing coding blocks according to the structure of a quadtree and a multi-type tree on the basis of a coding tree unit to obtain coding blocks with different sizes, determining optimal coding parameters according to the value of current secret information to be hidden on the basis of preset coding parameters and the mapping rule of hidden information, embedding the current secret information to be hidden into the coding blocks with the preset sizes by using the optimal coding parameters when the coding blocks with the preset sizes are coded for the first time, and further not directly modifying quantized transformation coefficients to be transmitted to realize information hiding, but hiding the secret information into the coding blocks by selecting the optimal coding parameters in the coding process of the coding blocks, wherein the coding process of the coding blocks is to code by adopting modulated information, the condition that final coefficients are inconsistent with original coefficients cannot occur, so that adjacent areas cannot be influenced, and the video quality can be ensured.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.

Fig. 1 is a schematic diagram of information embedding positions corresponding to different intra prediction modes in a 4 × 4 luma block according to embodiment 1 of the present invention;

FIG. 2 is a schematic diagram of the positions of prediction pixels and reference pixels in a 4 × 4 prediction block provided in embodiment 1 of the present invention;

fig. 3 is a schematic position diagram of a prediction block and an adjacent coding block provided in embodiment 1 of the present invention;

fig. 4 is a flowchart of a hiding method according to embodiment 1 of the present invention;

fig. 5 is a schematic diagram illustrating a partition result of an encoding block according to embodiment 1 of the present invention;

fig. 6 is a schematic diagram of the secret information embedding positions of a luminance block and a chrominance block provided in embodiment 1 of the present invention;

FIG. 7 is a distribution diagram of selection probabilities of selecting other transforms excluding the optimal transform according to embodiment 1 of the present invention;

fig. 8 is a selection probability distribution diagram for selecting other chroma prediction modes excluding the optimal chroma prediction mode according to embodiment 1 of the present invention;

fig. 9 is a system block diagram of the concealment system provided in embodiment 2 of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The invention aims to provide a video information hiding method and a video information hiding system, which hide secret information by selecting optimal coding parameters in the process of coding a video, so that the video quality can be ensured while hiding the video information.

In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.

Example 1:

taking the h.265/HEVC coding standard as an example, a method for hiding information by modifying quantized transform coefficients based on the existing coding standard is introduced:

in the h.265/HEVC coding standard, there are four valid intra-prediction block sizes, ranging from 4 × 4 to 32 × 32. Due to the characteristics of the human visual system, video visual quality distortion caused by embedding information in smooth areas is more obvious, and detection is difficult in complex areas. Therefore, when the h.265/HEVC encodes a video, blocks with different sizes are selected for prediction according to the video content, generally, a block with a relatively large size is selected for prediction in a smooth area, and a small block is selected for prediction in a complex area. The smaller the block size, the less impact on the visual quality of the video. In addition, the amount of information is larger in the I frame than in the P frame and the B frame, and the number of luminance blocks is larger than that of chrominance blocks. Therefore, the 4 × 4 luminance block of the I frame of the intra prediction is generally selected as the embedding region.

h.265/HEVC uses a new tool to improve coding efficiency: pattern correlation coefficient scan (MDCS). For intra-coded blocks, the scanning order of coefficients in luma blocks of 4 × 4 and 8 × 8 sizes is determined by the intra prediction mode, which is a mapping relationship of the intra prediction mode and the coefficient scanning order, as shown in table 1. Vertical and horizontal scanning refer to the respective raster scan order, while diagonal scanning refers to diagonal scanning in the direction from top left to bottom right.

TABLE 1

Intra prediction mode	Coefficient scan order of 4 × 4 and 8 × 8 size
		Angle prediction mode (6-14)	Vertical scanning
Angle prediction mode (22-30)	Horizontal scanning
		Other modes	Diagonal scanning

When modifying the quantized transform coefficients QTC, the setting of the scanning order is related to the visual quality and compression efficiency of the video. In h.264/AVC, diagonal elements of the QTC matrix are easier to modify than other elements. However, in h.265/HEVC, the number of prediction modes is larger than in h.264/AVC, and especially the scan order of 4 × 4 and 8 × 8 luma blocks is determined by the intra prediction mode. Finally, reference is made to the MDCS concept defined in h.265/HEVC and the position of the QTC that has to vary with its prediction mode is chosen differently. Fig. 1 is a schematic diagram of information embedding positions corresponding to different intra prediction modes in a 4 × 4 luma block. FIG. 1 (a) is a schematic diagram of information embedding positions when the intra prediction modes are 6 to 14; FIG. 1 (b) is a schematic diagram of information embedding positions when the intra prediction modes are 22 to 30; fig. 1 (c) is a diagram illustrating the information embedding positions when the intra prediction mode is the other mode.

To prevent the variation of QTC from negatively affecting entropy coding, coefficients greater than 1 in the significance of the 4x4 luma block are typically modified. The existing information embedding process is as follows:

1. selecting a 4x4 luma block with significant bits in an intra prediction process;

2. in the selected block, for Q _k (i, j) > 1, if the intra prediction mode is 6-14, modify the QTC in the vertical direction: q _k (i, 0), i =0 to 3; if the intra prediction mode is 22-30, modify QTC in the horizontal direction: q _k (0,j), j =0 to 3; if the intra-prediction mode is the other mode, then modify the QTC for the diagonal direction. The modification rules are as follows:

the modified QTC value of the ith row and the jth column; q _k (i, j) is the QTC value of the ith row and the jth column; w is a _k Indicating the kth information to be embedded.

Intra-prediction is used in I-frames of h.265/HEVC to reduce spatial redundancy of video sequences. The inner pixels of the 4 × 4 prediction block are formed by prediction using boundary pixels of upper and left reference pixels of the neighboring block. As shown in FIG. 2, the neighboring reference pixels of the current prediction block include R _0,0 、R _1,0 To R _2N,0 、R _0,1 To R _0,2N . The final coding block can select one prediction mode with the best prediction effect from 33 angle, plane and DC prediction modes. After the prediction mode is selected, the prediction pixel point P may be generated by the reference pixel and the selected prediction mode _1,1 To P _N,N 。

Assuming current predictionThe block is B _i,j In intra prediction, B _i,j Each pixel point of (1) is the sum of the predicted value and the residual value. Since the predictor is calculated from the grey reference pixels and prediction mode in fig. 3, the neighboring block B is _i-1,j-1 、B _i,j-1 、B _i-1,j And B _i-1,j+1 Will propagate to B by intra prediction _i,j . As shown in fig. 3. This cumulative visual distortion from the upper left corner to the lower right corner is defined as intra-frame distortion drift.

The method can be known that, because the embedding operation in the prior art is realized by directly modifying the quantitative transform coefficient QTC, the final coefficient generates deviation compared with the original coefficient, then the adjacent areas are influenced by intra-frame prediction to generate prediction errors, and the errors are always transmitted by the intra-frame prediction, so that the intra-frame distortion drift phenomenon is gradually accumulated to cause visual distortion, the visual concealment of information hiding is damaged, and the video quality is deteriorated.

In order to solve the above problem, an object of the present embodiment is to provide a video information hiding method, as shown in fig. 4, the hiding method includes: in encoding each I frame of a video, the following steps are performed:

s1: dividing coding blocks according to the structure of a quadtree and a multi-type Tree on the basis of a Coding Tree Unit (CTU) to obtain coding blocks with different sizes;

as shown in fig. 5, it is a diagram illustrating the division result of the coding block. As can be seen from fig. 5, the divided coding blocks are not identical, but there are a plurality of coding blocks with different sizes.

S2: determining an optimal encoding parameter according to the value of the current secret information to be hidden based on a preset encoding parameter and a mapping rule of the hidden information; the current secret information to be hidden is secret information corresponding to a coding block with a preset size;

in this embodiment, the coding block with the predetermined size may be a 4 × 4 coding block, and the 4 × 4 coding block is selected as an embedded carrier for information hiding. The embodiment only carries out the embedding of the secret information in the encoding process of the encoding blocks with the preset size, each encoding block with the preset size corresponds to one bit of secret information in the information to be embedded, and the embedding of the secret information is carried out through the encoding process of the encoding blocks with the preset size. The dividing process is synchronized with the embedding process, and when the encoding block with the preset size is divided, the secret information is embedded aiming at the encoding block with the preset size.

S3: and when the coding block with the preset size is coded for the first time, embedding the current secret information to be hidden into the coding block with the preset size by utilizing the optimal coding parameters.

The hiding method of the embodiment does not directly modify the quantized transform coefficient to be transmitted, but hides the secret information into the coding block by selecting the optimal coding parameter in the coding process of the coding block, and the coding process of the coding block is to adopt the modulated information for coding, so that the situation that the final coefficient is inconsistent with the original coefficient can not occur, the adjacent area can not be influenced, and the video quality can be ensured.

The H.266/VVC is a coding standard formally released in 7 months in 2020, still adopts a hybrid coding framework, still comprises a plurality of modules of prediction, transformation, quantization and entropy coding, but a plurality of coding tools are additionally arranged in each module to improve the compression efficiency. For Intra Prediction, h.266/VVC introduces several new mechanisms, such as Matrix Weighted Intra Prediction (MIP), multiple Reference Line (MRL), and Cross-Component Linear Prediction Model (CCLM), which increases the flexibility of mode selection and is therefore suitable for information hiding. Furthermore, h.266/VVC designs a novel adaptive transform selection mechanism called Multiple transform kernel selection (MTS), where horizontal and vertical residuals can be selected from different DCT or DST transform kernels, and this hybrid DCT + DST scheme based selection makes compression more efficient, which provides further flexibility for information hiding.

As an alternative implementation, this embodiment proposes a new h.266/VVC compressed video information hiding method by exploring h.266/VVC unique intra coding tools, which achieves a better tradeoff between hiding capacity, video quality and compression efficiency. Based on this, the coding block in the above hiding method may be a luminance block, and the coding parameter is MTS transform; or, the coding block is a chroma block, and the coding parameter is a chroma prediction mode; or, the coding blocks are a luminance block and a chrominance block, and the coding parameters are MTS transformation and chrominance prediction modes.

1. When the coding block is a luminance block and the coding parameter is MTS transform, as shown in fig. 6, in the process of coding each I frame of the video, the following steps are performed:

1. on the basis of the coding tree unit, dividing the brightness blocks according to the structures of the quadtree and the multi-type tree to obtain the brightness blocks with different sizes;

2. determining the optimal transformation according to the value of the current secret information to be hidden based on the preset MTS transformation and the mapping rule of the hidden information; the current secret information to be hidden is secret information corresponding to a brightness block with a preset size;

before determining the optimal transformation, the concealment method of the present embodiment further includes: and dividing the MTS transformation set into two transformation groups based on a grouping strategy of probability statistical analysis, wherein the value of the secret information corresponding to one transformation group is 1, and the value of the secret information corresponding to the other transformation group is 0, so as to obtain a preset MTS transformation and hidden information mapping rule. The MTS transform set includes a plurality of transforms, and specifically, the h.266/VVC encodes luma blocks using the MTS scheme and introduces two new transform matrices: DST-VII (DST 7) and DCT-VIII (DCT 8), which may apply different transformation matrices in the horizontal and vertical directions of the luminance block, the MTS transformation set of the present embodiment has six transformations as shown in table 2.

TABLE 2

Specifically, the MTS transform set includes transform 0 that uses DCT2 transform matrices in both the horizontal and vertical directions; the horizontal direction and the vertical direction adopt the transformation 1 of a TS transformation matrix; the horizontal direction and the vertical direction both adopt the transformation 2 of a DST7 transformation matrix; transform 3 of DCT8 transform matrix is adopted in horizontal direction and DST7 transform matrix is adopted in vertical direction; a DST7 transformation matrix is adopted in the horizontal direction, and a DCT8 transformation matrix is adopted in the vertical direction; both the horizontal and vertical directions use the transform 5 of the DCT8 transform matrix.

Here, since the hidden information is in a binary form, i.e., "0" and "1", the MTS transformation set may be divided into two transformation groups to map "0" and "1", respectively. For this reason, the present embodiment designs a grouping strategy based on probabilistic statistical analysis to minimize the negative impact on the video reconstruction quality when switching the optimal transform to another transform for information hiding. To mitigate the negative impact of concealment on the quality of the video reconstruction, the sub-optimal transform is defined as the most frequent transform selected by the h.266/VVC encoder when excluding the optimal transform, which is preferably divided into the opponent set of the optimal transform. Based on this, the grouping strategy based on the probabilistic statistical analysis specifically includes the following steps:

1) Determining suboptimal transformation corresponding to the optimal transformation by taking each transformation as the optimal transformation;

specifically, each transformation is used as an optimal transformation, the selection probability of selecting any of the other transformations when the optimal transformation is excluded by the encoder is determined, and the transformation with the highest selection probability is selected as the suboptimal transformation corresponding to the optimal transformation. As shown in fig. 7, which is a selection probability distribution graph for selecting other transforms when the optimal transform is excluded. For example, fig. 7 (a) shows the distribution of the probabilities for the remaining transforms when the optimal transform is 0, and it is found that transform 2 is selected much more frequently than the other transforms, so transform 2 is a sub-optimal transform for transform 0, and thus transform 0 and transform 2 are preferably in different transform groups.

2) And dividing the MTS transformation set into two transformation groups according to the principle that the optimal transformation and the suboptimal transformation are positioned in different groups.

Based on fig. 7, based on the principle that the optimal transformation and the suboptimal transformation are located in different groups, the transformation {2,3,5} is located in the first transformation group, the transformation {0,1,4} is located in the second transformation group, the two transformation groups respectively map "0" and "1" of the secret information, and then the MTS transformation set is divided into two transformation groups, so as to obtain the preset mapping rule of the MTS transformation and the hidden information.

Determining the optimal transformation according to the value of the current secret information to be hidden based on the preset MTS transformation and the mapping rule of the hidden information may include: based on a preset MTS transformation and a mapping rule of hidden information, selecting a transformation group according to the value of the current secret information to be hidden; an optimal transform is selected in the selected set of transforms using a first Lagrangian rate-distortion model.

Specifically, the value of the current secret information to be hidden is read, a transformation group is selected according to the preset MTS transformation and hidden information mapping rule and the value of the current secret information to be hidden, that is, the selection range of the transformation is determined, and if the value of the current secret information to be hidden is "1", the optimal transformation of the 4 × 4 luminance block is selected from the transformations {0,1,4} listed in table 2. Otherwise, the best transform for the 4x4 luma block is selected from the transforms {2,3,5} listed in table 2. The process of selecting the optimal transformation is selected by utilizing the first Lagrange rate distortion model, so that the video quality is not greatly influenced, and the formula of the first Lagrange rate distortion model is as follows:

in the formula (1), mtsIdx and mtsIdx _h Respectively representing the indices of the different transforms and the selected optimal transform for information hiding, D (-) representing the distortion caused by the transform using the mtsid x index, R (-) representing the number of coding bits required for the transform using the mtsid x index, and λ being the lagrangian parameter.

3. When a luminance block of a predetermined size is first encoded, the current secret information to be hidden is embedded in the luminance block of the predetermined size using an optimal transform. To avoid re-prediction and calculation of the residual, it is chosen to embed the information when the current luma block is first encoded.

2. When the coding block is a chroma block and the coding parameters are chroma prediction mode, as shown in fig. 6, in the process of coding each I frame of the video, the following steps are performed:

1. based on the coding tree unit, dividing the chrominance blocks according to the structure of the quadtree and the multi-type tree to obtain chrominance blocks with different sizes;

2. determining an optimal chroma prediction mode according to the value of current secret information to be hidden based on a preset chroma prediction mode and a mapping rule of hidden information; the current secret information to be hidden is secret information corresponding to a chroma block with a preset size;

before determining the optimal chroma prediction mode, the concealment method of this embodiment further includes: dividing a chroma prediction mode set into two chroma prediction mode groups based on a grouping strategy of probability statistical analysis, wherein the value of secret information corresponding to one chroma prediction mode group is 1, and the value of secret information corresponding to the other chroma prediction mode group is 0, so as to obtain a preset mapping rule of a chroma prediction mode and hidden information; the chroma prediction mode set includes a plurality of chroma prediction modes.

Similar to the mapping rule for obtaining the preset MTS transformation and hidden information, the chroma prediction mode set is divided into two different chroma prediction mode groups to map different secret information. The h.266/VVC employs 8 CHROMA prediction modes, including five conventional CHROMA prediction modes, namely PLANAR (CHROMA prediction mode 0), VER (CHROMA prediction mode 50), HOR (CHROMA prediction mode 18), DC (CHROMA prediction mode 1), DM CHROMA (CHROMA prediction mode 70), and three innovative cross-component linear models, including LM CHROMA (CHROMA prediction mode 67), MDLM L (CHROMA prediction mode 68), and MDLM T (CHROMA prediction mode 69), to reduce redundancy between color components.

The grouping strategy based on the probability statistical analysis specifically includes the following steps of:

1) Each chroma prediction mode is used as an optimal chroma prediction mode, and a suboptimal chroma prediction mode corresponding to the optimal chroma prediction mode is obtained;

specifically, each chroma prediction mode is used as an optimal chroma prediction mode, the selection probability of selecting any other chroma prediction mode when the optimal chroma prediction mode is eliminated by an encoder is determined, and the chroma prediction mode with the highest selection probability is selected as a suboptimal chroma prediction mode corresponding to the optimal chroma prediction mode. As shown in fig. 8, it is a selection probability distribution diagram for selecting other chroma prediction modes when excluding the optimal chroma prediction mode.

2) And dividing the chroma prediction mode set into two chroma prediction mode groups according to the principle that the optimal chroma prediction mode and the suboptimal chroma prediction mode are positioned in different groups.

Finally, by using this strategy, chroma prediction modes {0, 68, 69, 70} are located in the first set of chroma prediction modes, chroma prediction modes {1, 18, 50, 67} are located in the second set of chroma prediction modes, and the two sets of chroma prediction modes map secret information "0" and "1", respectively.

The determining the optimal chroma prediction mode according to the value of the current secret information to be hidden based on the preset chroma prediction mode and the mapping rule of the hidden information specifically comprises the following steps: based on a preset chroma prediction mode and a mapping rule of hidden information, selecting a chroma prediction mode group according to the value of current secret information to be hidden; selecting an optimal chroma prediction mode among the selected set of chroma prediction modes using a second Lagrangian rate-distortion model.

Reading the current secret information to be hidden, and appointing the selection range of the chroma prediction mode according to the preset mapping rule of the chroma prediction mode and the hidden information and the value of the current secret information to be hidden. Therefore, if the secret information is "1", the optimal chroma prediction mode in the 4 × 4 chroma block is selected from the chroma prediction modes {1, 18, 50, 67 }. Otherwise, it is selected from the

chroma prediction modes

0, 68, 69, 70. The selection process is based on a second lagrangian rate-distortion model, which is as follows:

in formula (2), mode and mode _h Respectively representing different chroma prediction modes and the selected optimal chroma prediction mode of information hiding, D (-) represents distortion caused by the mode chroma prediction mode, R (-) represents the number of coding bits required by using the mode chroma prediction mode, and lambda is a Lagrange parameter. By introducing a selection mechanism based on a Lagrange rate distortion model, the coding parameters are more optimal, and the influence on the video quality is smaller.

3. When a chroma block with a preset size is coded for the first time, the optimal chroma prediction mode is utilized to embed the current secret information to be hidden into the chroma block with the preset size.

3. When the coding block is a luminance block and a chrominance block, and the coding parameters are MTS transformation and chrominance prediction modes, the division and the coding of the luminance block and the chrominance block are mutually independent in an I frame, so that the luminance block can be divided and coded first, and then the chrominance block can be divided and coded; or the chrominance blocks are divided and encoded first and then the luminance blocks are divided and encoded. The manner of dividing and encoding the luminance block is, for example, the first manner, and the manner of dividing and encoding the chrominance block is, for example, the second manner, which are not described herein again.

When the luminance block and the chrominance block are divided and coded simultaneously, the secret information is hidden into the luminance block by modifying the selected MTS transformation, and the chrominance prediction mode is further selected to increase the hiding capacity and provide better video reconstruction quality. Compared with the prior art, the method has the advantages that the distortion drift phenomenon cannot be generated, the influence on the video quality is small, and the embedding capacity can be ensured.

The hiding method of the embodiment has the following advantages:

1. because the embedding operation in the prior art is to directly modify the quantized transform coefficients, the final coefficients generate deviation compared with the original coefficients, and then the adjacent regions are influenced through intra-frame prediction to generate prediction errors, and the errors are transmitted all the time through the intra-frame prediction, so that the intra-frame distortion drift phenomenon is gradually accumulated to cause visual distortion and destroy the visual concealment of information hiding. However, in this embodiment, the coefficients to be transmitted are not directly modified, but other coding characteristics in the coding process are utilized: transform, chroma prediction mode, and is a process of encoding: the secret information is hidden when the optimal chroma prediction mode is selected and the optimal transformation is carried out, the modulated information is adopted for coding in the current block coding process, the situation that the final coefficient is inconsistent with the original coefficient cannot occur, therefore, the adjacent area cannot be influenced, and the video quality cannot be influenced.

2. MTS is a unique coding tool in h.266/VVC to hide information into luma blocks. It has less impact on video quality and compression efficiency than other tools in luminance blocks, such as MIP and MRL. This feature ensures high video quality and sufficient concealment capacity with little impact on compression efficiency, which is not possible with any information concealment method designed for previous video coding standards, where the transform is fixed, since this tool cannot be used by previous coding standards. Another unique tool in h.266/VVC, CCLM, is also used to hide information into chroma blocks, which introduces several new chroma prediction modes. In this way, chroma prediction mode selection for information hiding is extended, resulting in better video quality and lower bit rate increments.

The embodiment provides an H.266/VVC video information hiding method based on luminance transformation and chroma prediction mode selection by selecting transformation and chroma prediction modes aiming at intra-frame distortion drift, which is the first information hiding method specially designed for H.266/VVC and realizes better information hiding performance compared with the hiding method designed for the previous compression standard. The specific implementation process is as follows:

1. and (3) embedding a brightness block: first, a 4 × 4 luminance block is selected, and secret information to be embedded is read. Secondly, in order to avoid re-prediction and calculation of the residual, secret information is embedded when the current luma block is first encoded, a selection range of transform is specified according to the secret information to be embedded, and then the best transform is selected from among them according to a rate-distortion cost. Thirdly, in order to correctly decode the optimal transform during decoding, when the embedded information is 1, it is required to check whether the CBF, which is the coding block flag of the current luma block, is 1. If the CBF of the current luminance block is 0, a flag is not sent to indicate a specific transform group, and the secret information corresponding to the default transform group is 0 during decoding, and the embedded secret information is erroneously judged to be 0. In order to avoid the case of information misjudgment, it is necessary to modify the transform quantization coefficients of the current luminance block so that they have a value other than 0 when the embedded secret information is 1, thereby making CBF 1. Fourthly, the current luminance block embedding is finished, and the next 4 × 4 luminance block is waited to be traversed until all 4 × 4 luminance blocks are encoded.

2. Chroma block embedding: firstly, selecting a 4x4 chroma block and reading the secret information to be embedded. Secondly, when the current chroma block is coded for the first time, the range of the chroma prediction mode is appointed according to the secret information to be embedded, and then the optimal prediction mode is selected according to the rate distortion cost. Third, the embedding is finished, waiting for the next 4 × 4 chroma block to be traversed.

To verify the feasibility and performance of the method, the method of this embodiment is implemented based on the VVC official standard reference encoder VTM6.0, the GOP size is 4, the structure is "IPPP", the Quantization Parameter (QP) is set to 27, lfnst =1, the test frame number is 40, and the remaining parameters are set to the VTM default configuration. For better evaluation of experimental effects, 4 standard video sequences of different resolutions and different texture complexities were used: blowingBubbeles _416x240_50, partyScene _832x480, vidoo 1_1280x720, and BQTerrace _1920x1080.

PSNR is a common index for evaluating video quality, and an analysis algorithm represents the influence on video quality based on PSNR difference before and after embedding information, and analyzes the influence on video bit rate according to bit rate increase before and after information, as shown in table 3.

TABLE 3

The experimental result shows that the information hiding algorithm provided by the embodiment has huge embedding capacity and has small influence on video quality and coding.

The embodiment particularly designs a novel information hiding algorithm with high embedding capacity and no distortion drift based on the H.266/VVC video coding standard. Information is hidden by modifying newly added coding parameters in the H.266/VVC in the video coding process, and multi-transform kernel selection (MTS) and cross-component linear model (CCLM) are two proprietary coding tools in the H.266/VVC standard, which are coding characteristics not possessed in the previous coding standard. The method comprises the following specific steps: firstly, selecting an I frame for hiding, selecting a 4x4 small block as an embedded carrier when a luminance block is coded, and embedding information into the luminance block by modifying the selection of a transformation core. MTS has less impact on compression quality and efficiency than other tools. Furthermore, since the luma and chroma blocks in the I-frame are encoded separately, the CCLM is further used to embed information into the chroma blocks to further expand the embedding capacity. The information hiding algorithm is applied to a VTM6.0 video encoder of reference software of H.266/VVC, so that the encoding efficiency and the video quality can be ensured while the high embedding capacity is ensured, and the method can be conveniently put into practical application.

Example 2:

this embodiment is used to provide a video information hiding system, as shown in fig. 9, where the hiding system includes:

the dividing module M1 is used for dividing the coding blocks according to the structure of the quadtree and the multi-type tree on the basis of the coding tree unit to obtain the coding blocks with different sizes;

the determining module M2 is used for determining the optimal coding parameter according to the value of the current secret information to be hidden based on the preset coding parameter and the mapping rule of the hidden information; the current secret information to be hidden is secret information corresponding to a coding block with a preset size;

and an embedding module M3, configured to, when the coding block with the predetermined size is coded for the first time, embed the current secret information to be hidden into the coding block with the predetermined size by using the optimal coding parameter.

The emphasis of each embodiment in the present specification is on the difference from the other embodiments, and the same and similar parts among the various embodiments may be referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.

The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims

1. A method for concealing video information, the method comprising:

in encoding each I-frame of a video, the following steps are performed:

when the coding block with the preset size is coded for the first time, the current secret information to be hidden is embedded into the coding block with the preset size by using the optimal coding parameters;

the coding block is a brightness block, and the coding parameter is MTS conversion;

or, the coding blocks are a luminance block and a chrominance block, and the coding parameters are an MTS transform and a chrominance prediction mode.

2. The concealment method according to claim 1, wherein when the encoded block is a luma block and the encoding parameter is MTS transform, the determining an optimal encoding parameter according to a value of current secret information to be concealed based on a preset encoding parameter and a mapping rule of concealment information is as follows: determining the optimal transformation according to the value of the current secret information to be hidden based on the preset MTS transformation and the mapping rule of the hidden information; before determining the optimal transformation according to the value of the current secret information to be hidden based on the preset mapping rule of the MTS transformation and the hidden information, the hiding method further comprises the following steps: dividing an MTS transformation set into two transformation groups based on a grouping strategy of probability statistical analysis, wherein the value of secret information corresponding to one transformation group is 1, and the value of secret information corresponding to the other transformation group is 0, so as to obtain the preset MTS transformation and hidden information mapping rule; the set of MTS transforms includes a plurality of transforms;

wherein the grouping strategy based on the probabilistic statistical analysis specifically includes the following steps:

determining suboptimal transformation corresponding to the optimal transformation by taking each transformation as the optimal transformation;

and dividing the MTS transformation set into two transformation groups according to the principle that the optimal transformation and the suboptimal transformation are positioned in different groups.

3. The concealment method according to claim 2, wherein said determining, with each of said transforms as an optimal transform, a sub-optimal transform corresponding to said optimal transform specifically comprises:

and taking each transformation as an optimal transformation, determining the selection probability of selecting any other transformation when the optimal transformation is excluded by the encoder, and selecting the transformation with the highest selection probability as the suboptimal transformation corresponding to the optimal transformation.

4. The hiding method according to claim 2, wherein the determining an optimal transformation according to the value of the current secret information to be hidden based on the preset MTS transformation and the mapping rule of the hidden information specifically comprises:

based on the preset MTS transformation and the mapping rule of the hidden information, selecting a transformation group according to the value of the current secret information to be hidden;

an optimal transform is selected in the selected set of transforms using a first Lagrangian rate-distortion model.

5. A video information hiding system, said hiding system comprising:

the determining module is used for determining the optimal encoding parameter according to the value of the current secret information to be hidden based on the preset encoding parameter and the mapping rule of the hidden information; the current secret information to be hidden is secret information corresponding to a coding block with a preset size;

the embedding module is used for embedding the current secret information to be hidden into the coding block with the preset size by utilizing the optimal coding parameter when the coding block with the preset size is coded for the first time;