Space domain image steganography method and system based on texture disturbance minimization

By introducing a strategy of minimizing texture perturbation in the image steganography method, using the embedding probability generation network and discriminator network to optimize the embedding process, the security problem of steganography analysis model detection in the existing technology is solved, and higher security is achieved Reduced sexuality and perturbation.

CN119027298AActive Publication Date: 2024-11-26NANJING UNIV OF INFORMATION SCI & TECH
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411509432.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-28
Publication Date
2024-11-26
Estimated Expiration
2044-10-28

AI Technical Summary

Technical Problem

Existing image steganography methods based on generative adversarial networks ignore the mutual influence between adjacent embedding points and the feature perturbation of embedding on texture areas, making the algorithm more easily detected by the deep steganography analysis model, limiting security.

Method used

Using the airspace image steganography method based on texture perturbation minimization, the network model and discriminator network model are generated through pre-established embedding probability, and combined with adjacent embedded pixel enhancement modules and texture consistency loss function, the generator group and discriminator group are optimized. Loss function to minimize the perturbation of the embedding to the image and improve safety.

Benefits of technology

It effectively reduces image perturbation and improves security. Through adaptive information embedding and integration of texture consistency loss functions, satisfactory security performance is achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119027298A_ABST
    Figure CN119027298A_ABST
Patent Text Reader

Abstract

The invention discloses a texture disturbance minimization-based spatial domain image steganography method, which relates to the technical field of secret communication, and can consider the mutual influence between adjacent embedded pixels through a designed adjacent embedded pixel enhancement module, so that the embedding directions in the same direction are more gathered, and the image steganography accuracy is improved. Disturbance to the image can be effectively reduced so as to improve safety, and cooperative confrontation loss is generated by utilizing the obtained cooperative value and comprehensive information such as the probability graph and gradient information. In an iterative update process of a generator set and a discriminator set, adjacent modifications may be synchronized using the obtained loss function, thereby improving security performance, a texture consistency loss is proposed that integrates pixel-level and image-level texture perturbations to limit the allocation of embedding costs within high texture regions, based on the generation of asymmetric embedding costs, the performance of the image-level texture perturbations is improved, and the performance of the image-level texture perturbations is improved. And the satisfactory safety performance is realized by adaptively minimizing texture disturbance caused by information embedding.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of secret communication technology, and in particular to a spatial domain image steganography method and system based on texture disturbance minimization. Background Art

[0002] Image steganography is an algorithm that embeds secret information into an image imperceptibly. Recently, steganography methods based on generative adversarial networks have attracted more attention due to their ability to automatically generate embedding probabilities and excellent security. However, in this area of ​​research, the mutual influence between neighboring embedded points and the feature perturbation of the embedding on the texture area have been ignored, making these algorithms more easily detected by deep steganalysis models that use texture features and representation learning for steganalysis, which inevitably limits the security of existing algorithms. Summary of the invention

[0003] In order to solve the deficiencies mentioned in the above background technology, the object of the present invention is to provide a spatial domain image steganography method and system based on texture perturbation minimization.

[0004] In a first aspect, the purpose of the present invention can be achieved by the following technical solution: a spatial domain image steganography method based on texture perturbation minimization, the method comprising the following steps: Receiving a grayscale carrier image and secret information, inputting the grayscale carrier image and secret information into a pre-established embedding probability generation network model, outputting an embedding change probability map, inputting the embedding change probability map into an embedding simulator to obtain an embedding change map, and then outputting a simulated secret image; The grayscale carrier image and the simulated encrypted image are input into the pre-established discriminator network model, and the image discrimination result is output. The adaptive binary cross entropy loss function and pixel-level gradient feedback information are calculated based on the discrimination result. The embedding change map is input into the pre-built neighboring embedding pixel enhancement module, and the collaborative value matrix is ​​output. The collaborative adversarial loss function is obtained by combining the pixel-level gradient feedback information, the collaborative value matrix and the embedding change probability map. The entropy constraint loss function is obtained according to the embedding change probability map combined with the pre-set embedding density ratio. According to the texture features of the grayscale carrier image and the simulated dense image, an image-level texture consistency loss function and a pixel-level texture consistency loss function are obtained, and the image-level texture consistency loss function and the pixel-level texture consistency loss function are combined with the collaborative value matrix and the embedding change probability map to obtain a texture consistency loss function; According to the collaborative adversarial loss function, texture consistency loss function and entropy constraint loss function, the total loss function of the generator group is obtained. The adaptive binary cross entropy loss function is used as the loss function of the discriminator group. The total loss function of the generator group and the loss function of the discriminator group are updated alternately. The optimization goal is to minimize the total loss function of the generator group and the loss function of the discriminator group, and the output is the embedding probability map of image steganography.

[0005] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the pre-established embedding probability generation network model includes: a plurality of convolution operation groups and a plurality of deconvolution operation groups, the plurality of convolution operation groups include: a first convolution operation group, a second convolution operation group, a third convolution operation group, a fourth convolution operation group, a fifth convolution operation group, a sixth convolution operation group, a seventh convolution operation group, an eighth convolution operation group, and a ninth convolution operation group, the plurality of deconvolution operation groups include a first deconvolution operation group, a second deconvolution operation group, a third deconvolution operation group, a fourth deconvolution operation group, a fifth deconvolution operation group, a sixth deconvolution operation group, a seventh deconvolution operation group, and an eighth deconvolution operation group; The output of the first convolution operation group and the output of the first deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the second convolution operation group and the output of the second deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the third convolution operation group and the output of the third deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fourth convolution operation group and the output of the fourth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fifth convolution operation group and the output of the fifth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the sixth convolution operation group and the output of the sixth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the seventh convolution operation group and the output of the seventh deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; Among them, the multiple convolution operation groups all include a convolution layer, an activation layer, a batch normalization layer, a convolution layer, an activation layer, and a batch normalization layer arranged in sequence; the multiple deconvolution operation groups all include a deconvolution layer, an activation layer, a batch normalization layer, a deconvolution layer, an activation layer, and a batch normalization layer arranged in sequence; The output of the generative network is an embedding change probability map.

[0006] In combination with the first aspect, in certain implementations of the first aspect, the method also includes: the pre-established discriminator network model is a combination of steganalysis network models based on SiaStegNet and SRNet, and the neighboring embedded pixel enhancement module includes a weight unit and a collaborative value matrix generation process.

[0007] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the weight unit W is expressed as:

[0008] in For weight units at index the value at the index; The process of generating the collaboration value matrix is ​​expressed as:

[0009] in, The collaboration value matrix is The value of the index, To embed the perturbation graph in The value of the index, u is the horizontal control parameter of the weight unit, and v is the vertical control parameter of the weight unit.

[0010] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the texture consistency loss function is as follows:

[0011]

[0012] in, , They are the texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator respectively. , They are the pixel-level texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator respectively. is the image-level texture consistency loss function. is a weight hyperparameter used to control the range of the loss function.

[0013] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the pixel-level texture loss is expressed as:

[0014]

[0015]

[0016]

[0017] and Represents the carrier image and encrypted images High-frequency features extracted by 30 high-pass filter kernels, Represents pixel-level texture differences, represents the kth filter kernel among the 30 high-pass filter kernels, is an indicator function, indicating The part greater than 0, , , Represent the collaboration value matrix, positive embedding probability, and negative embedding probability respectively. The value of the index, represents the gradient calculated by the steganalyzer group. The calculation process is:

[0018] in, Represents the binary cross entropy loss corresponding to the most discriminative steganalyzer at present, and the input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differentiation operation.

[0019] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the image-level texture loss is expressed as:

[0020]

[0021]

[0022] in, , Represent the image-level texture features of the carrier image and the encrypted image, Indicates that the feature is at index The value at Express Find the L1-norm, Represents the carrier image Texture features In the index The value at . Represents the carrier image Texture features In the index The value at . Represents the index of the channel.

[0023] The input of pixel-level texture loss is the carrier image and encrypted images , the output is pixel-level texture loss.

[0024] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the adaptive binary cross entropy loss function as follows:

[0025] The input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differentiation operation.

[0026] The collaborative adversarial loss function is as follows:

[0027] in, is an indicator function, indicating The part greater than 0, , , Represent the collaboration value matrix, positive embedding probability, and negative embedding probability respectively. The value of the index, represents the gradient calculated by the steganalyzer group, and Represent the height and length of the image respectively; The loss function of the generator group is as follows:

[0028] in, , , To control the parameters of each part’s weight, , Represent the total loss function of the positive and negative embedding probability generators, respectively. , Represent the texture consistency loss functions corresponding to the positive and negative embedding probability generators, respectively. represents the pre-set entropy constraint loss function, which is calculated as follows:

[0029] in Indicates the modification direction. Indicates the embedding rate.

[0030] In a second aspect, in order to achieve the above-mentioned object, the present invention discloses a spatial domain image steganography system based on texture perturbation minimization, comprising: A simulated secret-containing module is used to receive a grayscale carrier image and secret information, input the grayscale carrier image and secret information into a pre-established embedding probability generation network model, output an embedding change probability map, input the embedding change probability map into an embedding simulator to obtain an embedding change map, and then output a simulated secret-containing image; The image discrimination module is used to input the grayscale carrier image and the simulated secret image into the pre-established discriminator network model, output the image discrimination result, and calculate the adaptive binary cross entropy loss function and pixel-level gradient feedback information according to the discrimination result; The collaborative adversarial module is used to input the embedding change map into the pre-built neighboring embedding pixel enhancement module, and output the collaborative value matrix. The collaborative adversarial loss function is obtained by combining the pixel-level gradient feedback information, the collaborative value matrix and the embedding change probability map. The entropy constraint loss function is obtained according to the embedding change probability map combined with the pre-set embedding density ratio. A texture loss module is used to obtain an image-level texture consistency loss function and a pixel-level texture consistency loss function according to the texture features of the grayscale carrier image and the simulated dense image, and to obtain a texture consistency loss function by combining the image-level texture consistency loss function and the pixel-level texture consistency loss function with the collaboration value matrix and the embedding change probability map; The image steganography module is used to obtain the total loss function of the generator group based on the collaborative adversarial loss function, the texture consistency loss function and the entropy constraint loss function, and use the adaptive binary cross entropy loss function as the loss function of the discriminator group. The total loss function of the generator group and the loss function of the discriminator group are alternately updated to minimize the total loss function of the generator group and the loss function of the discriminator group as the optimization goal, and the embedded probability map of the image steganography is output.

[0031] In combination with the second aspect, in some implementations of the second aspect, the system further includes: the embedding probability generation network model pre-established by the simulation secret module includes: multiple convolution operation groups and multiple deconvolution operation groups, the multiple convolution operation groups include: a first convolution operation group, a second convolution operation group, a third convolution operation group, a fourth convolution operation group, a fifth convolution operation group, a sixth convolution operation group, a seventh convolution operation group, an eighth convolution operation group, and a ninth convolution operation group, the multiple deconvolution operation groups include a first deconvolution operation group, a second deconvolution operation group, a third deconvolution operation group, a fourth deconvolution operation group, a fifth deconvolution operation group, a sixth deconvolution operation group, a seventh deconvolution operation group, and an eighth deconvolution operation group; The output of the first convolution operation group and the output of the first deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the second convolution operation group and the output of the second deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the third convolution operation group and the output of the third deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fourth convolution operation group and the output of the fourth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fifth convolution operation group and the output of the fifth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the sixth convolution operation group and the output of the sixth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the seventh convolution operation group and the output of the seventh deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; Among them, the multiple convolution operation groups all include a convolution layer, an activation layer, a batch normalization layer, a convolution layer, an activation layer, and a batch normalization layer arranged in sequence; the multiple deconvolution operation groups all include a deconvolution layer, an activation layer, a batch normalization layer, a deconvolution layer, an activation layer, and a batch normalization layer arranged in sequence; The output of the generative network is an embedding change probability map; The discriminator network model pre-established in the image discrimination module is a combination of steganalysis network models composed of SiaStegNet and SRNet. The neighboring embedded pixel enhancement module includes a weight unit and a collaborative value matrix generation process. Among them, the image discrimination module weight unit It is expressed as:

[0032] in For weight units at index The value at index.

[0033] The process of generating the collaboration value matrix is ​​expressed as:

[0034] in, The collaboration value matrix is The value of the index, To embed the perturbation graph in The value of the index, u is the horizontal control parameter of the weight unit, and v is the vertical control parameter of the weight unit; The texture consistency loss function of the texture loss module is as follows:

[0035]

[0036] in, , They are the texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator respectively. , They are the pixel-level texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator respectively. is the image-level texture consistency loss function. is a weight hyperparameter used to control the range of the loss function.

[0037] The pixel-level texture loss is expressed as:

[0038]

[0039]

[0040]

[0041] and Represents the carrier image and encrypted images High-frequency features extracted by 30 high-pass filter kernels, Represents pixel-level texture differences, represents the kth filter kernel among the 30 high-pass filter kernels, is an indicator function, indicating The part greater than 0, , , Represent the collaboration value matrix, positive embedding probability, and negative embedding probability respectively. The value of the index, represents the gradient calculated by the steganalyzer group. The calculation process is:

[0042] in, Represents the binary cross entropy loss corresponding to the most discriminative steganalyzer at present, and the input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differential operation; The image-level texture loss is expressed as:

[0043]

[0044]

[0045] in, , Represent the image-level texture features of the carrier image and the encrypted image, Indicates that the feature is at index The value at Express Find the L1-norm, Represents the carrier image Texture features In the index The value at . Represents the carrier image Texture features In the index The value at . Represents the index of the channel.

[0046] The input of pixel-level texture loss is the carrier image and encrypted images , the output is pixel-level texture loss; Adaptive binary cross entropy loss function for image steganography module as follows:

[0047] The input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differentiation operation.

[0048] The collaborative adversarial loss function is as follows:

[0049]

[0050] in, is an indicator function, indicating The part greater than 0, , , Represent the collaboration value matrix, positive embedding probability, and negative embedding probability respectively. The value of the index, represents the gradient calculated by the steganalyzer group, and Represent the height and length of the image respectively; The loss function of the generator group is as follows:

[0051] in, , , To control the parameters of each part’s weight, , Represent the total loss function of the positive and negative embedding probability generators, respectively. , Represent the texture consistency loss functions corresponding to the positive and negative embedding probability generators, respectively. represents the pre-set entropy constraint loss function, which is calculated as follows:

[0052] in Indicates the modification direction. Indicates the embedding rate.

[0053] Beneficial effects of the present invention: The neighboring embedded pixel enhancement module designed by the present invention can consider the mutual influence between neighboring embedded pixels, so that the embedding directions of the same direction are more concentrated, which can effectively reduce the disturbance to the image to improve security, and generate collaborative adversarial loss using the obtained collaborative value and comprehensive information such as probability map and gradient information. In the iterative update process of the generator set and the discriminator set, the obtained loss function can be used to synchronize adjacent modifications to improve security performance, and a texture consistency loss is proposed. It integrates pixel-level and image-level texture perturbations to limit the distribution of embedding costs in high-texture areas, and achieves satisfactory security performance by adaptively minimizing the texture perturbations caused by information embedding based on the generation of asymmetric embedding costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art description are briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative work. Figure 1 It is a schematic flow chart of the method of the present invention; Figure 2 is a schematic diagram of a neighboring embedded pixel enhancement module of the present invention; Figure 3 is a schematic diagram of a texture consistency loss module of the present invention; Figure 4 It is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] Embodiment 1: The following is an introduction to the relevant terms involved in the embodiments of the present application: Image steganography is a technique for hiding information in images, aiming to protect the transmission and storage security of information. It is different from cryptography because steganography hides the transmission process of information rather than the content of the information itself. Image steganography can be divided into two categories: robust steganography and fragile steganography, as well as reversible steganography and irreversible steganography. In addition, there is also traditional steganography and deep learning steganography.

[0057] The loss function or cost function is a function that maps the value of a random event or its related random variables to a non-negative real number to represent the "risk" or "loss" of the random event. In applications, the loss function is usually associated with the optimization problem as a learning criterion, that is, the model is solved and evaluated by minimizing the loss function.

[0058] like Figure 1 As shown, a spatial domain image steganography method based on texture perturbation minimization comprises the following steps: S101: receiving a grayscale carrier image and secret information, inputting the grayscale carrier image and secret information into a pre-established embedding probability generation network model, outputting an embedding modification probability map, inputting the embedding modification probability map into an embedding simulator, obtaining an embedding modification map, and then outputting a simulated secret image; Wherein, the pre-established embedding probability generation network model includes: a plurality of convolution operation groups and a plurality of deconvolution operation groups, the plurality of convolution operation groups include: a first convolution operation group, a second convolution operation group, a third convolution operation group, a fourth convolution operation group, a fifth convolution operation group, a sixth convolution operation group, a seventh convolution operation group, an eighth convolution operation group, and a ninth convolution operation group, and the plurality of deconvolution operation groups include a first deconvolution operation group, a second deconvolution operation group, a third deconvolution operation group, a fourth deconvolution operation group, a fifth deconvolution operation group, a sixth deconvolution operation group, a seventh deconvolution operation group, and an eighth deconvolution operation group; The output of the first convolution operation group and the output of the first deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the second convolution operation group and the output of the second deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the third convolution operation group and the output of the third deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fourth convolution operation group and the output of the fourth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fifth convolution operation group and the output of the fifth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the sixth convolution operation group and the output of the sixth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the seventh convolution operation group and the output of the seventh deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; Among them, the multiple convolution operation groups all include a convolution layer, an activation layer, a batch normalization layer, a convolution layer, an activation layer, and a batch normalization layer arranged in sequence; the multiple deconvolution operation groups all include a deconvolution layer, an activation layer, a batch normalization layer, a deconvolution layer, an activation layer, and a batch normalization layer arranged in sequence; The output of the generative network is an embedding change probability map.

[0059] S102: Inputting the grayscale carrier image and the simulated encrypted image into a pre-established discriminator network model, outputting an image discrimination result, and calculating an adaptive binary cross entropy loss function and pixel-level gradient feedback information according to the discrimination result; The pre-established discriminator network model is a combination of steganalysis network models based on SiaStegNet and SRNet, and the neighboring embedded pixel enhancement module includes a weight unit and a collaborative value matrix generation process; Weight Unit It is expressed as:

[0060] in For weight units at index The value at index.

[0061] like Figure 2 As shown, the collaborative value matrix generation process is expressed as:

[0062] in, The collaboration value matrix is The value of the index, To embed the perturbation graph in The value of the index, u is the horizontal control parameter of the weight unit, and v is the vertical control parameter of the weight unit; S103: input the embedding change map into a pre-built neighboring embedding pixel enhancement module, output a collaborative value matrix, combine pixel-level gradient feedback information, the collaborative value matrix and the embedding change probability map to obtain a collaborative adversarial loss function, and obtain an entropy constraint loss function according to the embedding change probability map combined with a pre-set embedding density ratio; S104: Figure 3 As shown, according to the texture features of the grayscale carrier image and the simulated dense image, the image-level texture consistency loss function and the pixel-level texture consistency loss function are obtained, and the image-level texture consistency loss function and the pixel-level texture consistency loss function are combined with the collaborative value matrix and the embedding change probability map to obtain the texture consistency loss function; The texture consistency loss function is as follows:

[0063]

[0064] in, , They are the texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator respectively. , They are the pixel-level texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator respectively. is the image-level texture consistency loss function. is a weight hyperparameter used to control the range of the loss function.

[0065] The pixel-level texture loss is expressed as:

[0066]

[0067]

[0068]

[0069] and Represents the carrier image and encrypted images High-frequency features extracted by 30 high-pass filter kernels, Represents pixel-level texture differences, represents the kth filter kernel among the 30 high-pass filter kernels, is an indicator function, indicating The part greater than 0, , , Represent the collaboration value matrix, positive embedding probability, and negative embedding probability respectively. The value of the index, represents the gradient calculated by the steganalyzer group. The calculation process is:

[0070] in, Represents the binary cross entropy loss corresponding to the most discriminative steganalyzer at present, and the input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differential operation; The pixel-level texture loss is expressed as:

[0071]

[0072]

[0073] in, , Represent the image-level texture features of the carrier image and the encrypted image, Indicates that the feature is at index The value at Express Find the L1-norm, Represents the carrier image Texture features In the index The value at . Represents the carrier image Texture features In the index The value at . Represents the index of the channel.

[0074] The input of pixel-level texture loss is the carrier image and encrypted images , the output is pixel-level texture loss.

[0075] S105: The total loss function of the generator group is obtained according to the collaborative adversarial loss function and the texture consistency loss function, and the adaptive binary cross entropy loss function is used as the loss function of the discriminator group. The total loss function of the generator group and the loss function of the discriminator group are updated alternately, and the minimization of the total loss function of the generator group and the loss function of the discriminator group is taken as the optimization goal, and the image steganography result is obtained as output.

[0076] In actual embedding, the generated probability map is converted into the embedding cost through the conversion formula, and the secret information is embedded and extracted using the pre-prepared steganographic coding such as STC. The above-mentioned conversion formula is as follows:

[0077] Adaptive Binary Cross Entropy Loss Function as follows:

[0078] The input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differentiation operation.

[0079] The collaborative adversarial loss function is as follows:

[0080]

[0081] in, is an indicator function, indicating The part greater than 0, , , Represent the collaboration value matrix, positive embedding probability, and negative embedding probability respectively. The value of the index, Represents the calculated gradient of the steganalyzer group. and Represents the height and length of the image respectively.

[0082] The loss function of the generator group is as follows:

[0083] in, , , is the parameter that controls the weight of each part. , Represent the total loss function of the positive and negative embedding probability generators, respectively. , Represent the texture consistency loss functions corresponding to the positive and negative embedding probability generators respectively. represents the pre-set entropy constraint loss function, which is calculated as follows:

[0084] in Indicates possible modification directions. Indicates the embedding rate, and its unit is bpp (bit per pixel).

[0085] The performance of the steganography algorithm is evaluated by the detection error rate of the steganalyzer. ,in represents the false positive rate, Represents the false alarm rate.

[0086] An experimental comparison was performed on 0.1bpp to 0.4bpp.

[0087] On the BossBase dataset, the results are shown in Table 1: Table 1

[0088] On the Bows2 dataset, the results are shown in Table 2: Table 2

[0089] Experiments on multiple data sets show that the algorithm has high and outstanding security performance.

[0090] Embodiment 2: In the second aspect, as Figure 4 As shown, in order to achieve the above-mentioned purpose, the present invention discloses a spatial domain image steganography system based on texture perturbation minimization, comprising: The simulation secret module 11 is used to receive the grayscale carrier image and the secret information, input the grayscale carrier image and the secret information into a pre-established embedding probability generation network model, output an embedding change probability map, input the embedding change probability map into an embedding simulator to obtain an embedding change map, and then output a simulated secret image; The image discrimination module 12 is used to input the grayscale carrier image and the simulated encrypted image into a pre-established discriminator network model, output the image discrimination result, and calculate the adaptive binary cross entropy loss function and pixel-level gradient feedback information according to the discrimination result; The collaborative adversarial module 13 is used to input the embedding change map into a pre-built neighboring embedded pixel enhancement module, output a collaborative value matrix, combine the pixel-level gradient feedback information, the collaborative value matrix and the embedding change probability map to obtain a collaborative adversarial loss function, and obtain an entropy constraint loss function based on the embedding change probability map combined with a pre-set embedding density ratio; A texture loss module 14 is used to obtain an image-level texture consistency loss function and a pixel-level texture consistency loss function according to texture features of the grayscale carrier image and the simulated dense image, and to obtain a texture consistency loss function by combining the image-level texture consistency loss function and the pixel-level texture consistency loss function with the collaboration value matrix and the embedding change probability map; The image steganography module 15 is used to obtain the total loss function of the generator group according to the collaborative adversarial loss function, the texture consistency loss function and the entropy constraint loss function, use the adaptive binary cross entropy loss function as the loss function of the discriminator group, alternately update the total loss function of the generator group and the loss function of the discriminator group, and minimize the total loss function of the generator group and the loss function of the discriminator group as the optimization goal, and output the embedding probability map of the image steganography.

[0091] In combination with the second aspect, in some implementations of the second aspect, the system further includes: the embedding probability generation network model pre-established by the simulation encryption module 11 includes: multiple convolution operation groups and multiple deconvolution operation groups, the multiple convolution operation groups include: a first convolution operation group, a second convolution operation group, a third convolution operation group, a fourth convolution operation group, a fifth convolution operation group, a sixth convolution operation group, a seventh convolution operation group, an eighth convolution operation group, and a ninth convolution operation group, the multiple deconvolution operation groups include a first deconvolution operation group, a second deconvolution operation group, a third deconvolution operation group, a fourth deconvolution operation group, a fifth deconvolution operation group, a sixth deconvolution operation group, a seventh deconvolution operation group, and an eighth deconvolution operation group; The output of the first convolution operation group and the output of the first deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the second convolution operation group and the output of the second deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the third convolution operation group and the output of the third deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fourth convolution operation group and the output of the fourth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fifth convolution operation group and the output of the fifth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the sixth convolution operation group and the output of the sixth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the seventh convolution operation group and the output of the seventh deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; Among them, the multiple convolution operation groups all include a convolution layer, an activation layer, a batch normalization layer, a convolution layer, an activation layer, and a batch normalization layer arranged in sequence; the multiple deconvolution operation groups all include a deconvolution layer, an activation layer, a batch normalization layer, a deconvolution layer, an activation layer, and a batch normalization layer arranged in sequence; The output of the generative network is an embedding change probability map; The discriminator network model pre-established by the image discrimination module 12 is a combination of steganalysis network models based on SiaStegNet and SRNet, and the neighboring embedded pixel enhancement module includes a weight unit and a collaborative value matrix generation process; Among them, the image discrimination module 12 weight unit It is expressed as:

[0092] in For weight units at index The value at index.

[0093] The process of generating the collaboration value matrix is ​​expressed as:

[0094] in, The collaboration value matrix is The value of the index, To embed the perturbation graph in The value of the index, u is the horizontal control parameter of the weight unit, and v is the vertical control parameter of the weight unit; The texture consistency loss function of texture loss module 15 is as follows:

[0095]

[0096] in, , They are the texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator respectively. , They are the pixel-level texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator respectively. is the image-level texture consistency loss function. is a weight hyperparameter used to control the range of the loss function.

[0097] The pixel-level texture loss is expressed as:

[0098]

[0099]

[0100]

[0101] and Represents the carrier image and encrypted images High-frequency features extracted by 30 high-pass filter kernels, Represents pixel-level texture differences, represents the kth filter kernel among the 30 high-pass filter kernels, is an indicator function, indicating The part greater than 0, , , Represent the collaboration value matrix, positive embedding probability, and negative embedding probability respectively. The value of the index, represents the gradient calculated by the steganalyzer group. The calculation process is:

[0102] in, Represents the binary cross entropy loss corresponding to the most discriminative steganalyzer at present, and the input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differential operation; The image-level texture loss is expressed as:

[0103]

[0104]

[0105] in, , Represent the image-level texture features of the carrier image and the encrypted image, Indicates that the feature is at index The value at Express Find the L1-norm, Represents the carrier image Texture features In the index The value at . Represents the carrier image Texture features In the index The value at . Represents the index of the channel.

[0106] The input of pixel-level texture loss is the carrier image and encrypted images , the output is pixel-level texture loss; Adaptive binary cross entropy loss function for image steganography module as follows:

[0107] The input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differentiation operation.

[0108] The collaborative adversarial loss function is as follows:

[0109]

[0110] in, is an indicator function, indicating The part greater than 0, , , Respectively represent the collaboration value matrix, the positive embedding probability, and the negative embedding probability in The value of the index, Represents the calculated gradient of the steganalyzer group. and Represents the height and length of the image respectively.

[0111] The loss function of the generator group is as follows:

[0112] in, , , is the parameter that controls the weight of each part. , Represent the total loss function of the positive and negative embedding probability generators, respectively. , Represent the texture consistency loss functions corresponding to the positive and negative embedding probability generators respectively. represents the pre-set entropy constraint loss function, which is calculated as follows:

[0113] in Indicates possible modification directions. Indicates the embedding rate, and its unit is bpp (bit per pixel).

[0114] Based on the same inventive concept, the present invention also provides a computer device, which includes: one or more processors, and a memory for storing one or more computer programs; the program includes program instructions, and the processor is used to execute the program instructions stored in the memory. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is used to implement one or more instructions, specifically for loading and executing one or more instructions in a computer storage medium to implement the above method.

[0115] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium, on which a computer program is stored, and the computer program is executed by the processor to execute the above method. The storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electrical, magnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0116] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0117] The above shows and describes the basic principles, main features and advantages of the present disclosure. Those skilled in the art should understand that the present disclosure is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present disclosure. Without departing from the spirit and scope of the present disclosure, the present disclosure may have various changes and improvements, and these changes and improvements fall within the scope of the present disclosure to be protected.

Claims

1. A spatial domain image steganography method based on texture perturbation minimization, characterized in that: The method comprises the following steps: Receiving a grayscale carrier image and secret information, inputting the grayscale carrier image and secret information into a pre-established embedding probability generation network model, outputting an embedding change probability map, inputting the embedding change probability map into an embedding simulator to obtain an embedding change map, and then outputting a simulated secret image; The grayscale carrier image and the simulated encrypted image are input into the pre-established discriminator network model, and the image discrimination result is output. The adaptive binary cross entropy loss function and pixel-level gradient feedback information are calculated based on the discrimination result. The embedding change map is input into the pre-built neighboring embedding pixel enhancement module, and the collaborative value matrix is ​​output. The collaborative adversarial loss function is obtained by combining the pixel-level gradient feedback information, the collaborative value matrix and the embedding change probability map. The entropy constraint loss function is obtained according to the embedding change probability map combined with the pre-set embedding density ratio. According to the texture features of the grayscale carrier image and the simulated dense image, an image-level texture consistency loss function and a pixel-level texture consistency loss function are obtained, and the image-level texture consistency loss function and the pixel-level texture consistency loss function are combined with the collaborative value matrix and the embedding change probability map to obtain a texture consistency loss function; According to the collaborative adversarial loss function, texture consistency loss function and entropy constraint loss function, the total loss function of the generator group is obtained. The adaptive binary cross entropy loss function is used as the loss function of the discriminator group. The total loss function of the generator group and the loss function of the discriminator group are updated alternately. The optimization goal is to minimize the total loss function of the generator group and the loss function of the discriminator group, and the output is the embedding probability map of image steganography.

2. According to claim 1, a spatial domain image steganography method based on texture perturbation minimization is characterized in that: The pre-established embedding probability generation network model includes: a plurality of convolution operation groups and a plurality of deconvolution operation groups, the plurality of convolution operation groups include: a first convolution operation group, a second convolution operation group, a third convolution operation group, a fourth convolution operation group, a fifth convolution operation group, a sixth convolution operation group, a seventh convolution operation group, an eighth convolution operation group, and a ninth convolution operation group, the plurality of deconvolution operation groups include a first deconvolution operation group, a second deconvolution operation group, a third deconvolution operation group, a fourth deconvolution operation group, a fifth deconvolution operation group, a sixth deconvolution operation group, a seventh deconvolution operation group, and an eighth deconvolution operation group; The output of the first convolution operation group and the output of the first deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the second convolution operation group and the output of the second deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the third convolution operation group and the output of the third deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fourth convolution operation group and the output of the fourth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fifth convolution operation group and the output of the fifth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the sixth convolution operation group and the output of the sixth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the seventh convolution operation group and the output of the seventh deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; Among them, the multiple convolution operation groups all include a convolution layer, an activation layer, a batch normalization layer, a convolution layer, an activation layer, and a batch normalization layer arranged in sequence; the multiple deconvolution operation groups all include a deconvolution layer, an activation layer, a batch normalization layer, a deconvolution layer, an activation layer, and a batch normalization layer arranged in sequence; The output of the generative network is an embedding change probability map.

3. According to the spatial domain image steganography method based on texture perturbation minimization according to claim 1, it is characterized in that: The pre-established discriminator network model is a combination of steganalysis network models based on SiaStegNet and SRNet, and the neighboring embedded pixel enhancement module includes a weight unit and a collaborative value matrix generation process.

4. According to claim 3, a spatial domain image steganography method based on texture perturbation minimization is characterized in that: The weight unit It is expressed as: in For weight units at index the value at the index; The process of generating the collaboration value matrix is ​​expressed as: in, The collaboration value matrix is the value at the index, To embed the perturbation graph in The value of the index, u is the horizontal control parameter of the weight unit, and v is the vertical control parameter of the weight unit.

5. The spatial domain image steganography method based on texture perturbation minimization according to claim 1, characterized in that: The texture consistency loss function is as follows: in, , are the texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator, respectively. , are the pixel-level texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator, respectively. is the image-level texture consistency loss function, is the weight hyperparameter.

6. The spatial domain image steganography method based on texture perturbation minimization according to claim 1, characterized in that: The pixel-level texture consistency loss function is expressed as: and Represents the carrier image and encrypted images High-frequency features extracted by 30 high-pass filter kernels, Represents pixel-level texture differences, represents the kth filter kernel among the 30 high-pass filter kernels, is an indicator function, indicating The part greater than 0, , , Respectively represent the collaboration value matrix, the positive embedding probability, and the negative embedding probability in The value of the index, represents the gradient calculated by the steganalyzer group. The calculation process is: in, Represents the binary cross entropy loss corresponding to the most discriminative steganalyzer at present, and the input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differentiation operation.

7. The spatial domain image steganography method based on texture perturbation minimization according to claim 1, characterized in that: The image-level texture consistency loss function is expressed as: in, , Represent the image-level texture features of the carrier image and the encrypted image, Indicates that the feature is at index The value at Express Find the L1-norm, Represents the carrier image Texture features In the index The value at Represents the carrier image Texture features In the index The value at Represents the index of the channel; The input of pixel-level texture loss is the carrier image and encrypted images , the output is pixel-level texture loss.

8. The spatial domain image steganography method based on texture perturbation minimization according to claim 1, characterized in that: The adaptive binary cross entropy loss function as follows: The input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differential operation; The collaborative adversarial loss function is as follows: in, is an indicator function, indicating The part greater than 0, , , Respectively represent the collaboration value matrix, the positive embedding probability, and the negative embedding probability in The value of the index, represents the gradient calculated by the steganalyzer group, and Represent the height and length of the image respectively; The loss function of the generator group is as follows: in, , , To control the parameters of each part’s weight, , Represent the total loss function of the positive and negative embedding probability generators, respectively. , Represent the texture consistency loss functions corresponding to the positive and negative embedding probability generators, respectively. represents the pre-set entropy constraint loss function, which is calculated as follows: in Indicates the modification direction. Indicates the embedding rate.

9. A spatial domain image steganography system based on texture perturbation minimization, characterized in that: include: A simulated secret-containing module is used to receive a grayscale carrier image and secret information, input the grayscale carrier image and secret information into a pre-established embedding probability generation network model, output an embedding change probability map, input the embedding change probability map into an embedding simulator to obtain an embedding change map, and then output a simulated secret-containing image; The image discrimination module is used to input the grayscale carrier image and the simulated secret image into the pre-established discriminator network model, output the image discrimination result, and calculate the adaptive binary cross entropy loss function and pixel-level gradient feedback information according to the discrimination result; The collaborative adversarial module is used to input the embedding change map into the pre-built neighboring embedding pixel enhancement module, and output the collaborative value matrix. The collaborative adversarial loss function is obtained by combining the pixel-level gradient feedback information, the collaborative value matrix and the embedding change probability map. The entropy constraint loss function is obtained according to the embedding change probability map combined with the pre-set embedding density ratio. A texture loss module is used to obtain an image-level texture consistency loss function and a pixel-level texture consistency loss function according to the texture features of the grayscale carrier image and the simulated dense image, and to obtain a texture consistency loss function by combining the image-level texture consistency loss function and the pixel-level texture consistency loss function with the collaboration value matrix and the embedding change probability map; The image steganography module is used to obtain the total loss function of the generator group based on the collaborative adversarial loss function, the texture consistency loss function and the entropy constraint loss function, and use the adaptive binary cross entropy loss function as the loss function of the discriminator group. The total loss function of the generator group and the loss function of the discriminator group are alternately updated to minimize the total loss function of the generator group and the loss function of the discriminator group as the optimization goal, and the embedded probability map of the image steganography is output.

10. The spatial domain image steganography system based on texture perturbation minimization according to claim 9, characterized in that: The pre-established embedding probability generation network model of the simulated secret module includes: a plurality of convolution operation groups and a plurality of deconvolution operation groups, the plurality of convolution operation groups include: a first convolution operation group, a second convolution operation group, a third convolution operation group, a fourth convolution operation group, a fifth convolution operation group, a sixth convolution operation group, a seventh convolution operation group, an eighth convolution operation group, and a ninth convolution operation group, the plurality of deconvolution operation groups include a first deconvolution operation group, a second deconvolution operation group, a third deconvolution operation group, a fourth deconvolution operation group, a fifth deconvolution operation group, a sixth deconvolution operation group, a seventh deconvolution operation group, and an eighth deconvolution operation group; The output of the first convolution operation group and the output of the first deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the second convolution operation group and the output of the second deconvolution operation group are spliced ​​on the channel and then input into the deconvolution first operation group to form a jump structure; The output of the third convolution operation group and the output of the third deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fourth convolution operation group and the output of the fourth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the fifth convolution operation group and the output of the fifth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the sixth convolution operation group and the output of the sixth deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; The output of the seventh convolution operation group and the output of the seventh deconvolution operation group are spliced ​​on the channel and then input into the first deconvolution operation group to form a jump structure; Among them, the multiple convolution operation groups all include a convolution layer, an activation layer, a batch normalization layer, a convolution layer, an activation layer, and a batch normalization layer arranged in sequence; the multiple deconvolution operation groups all include a deconvolution layer, an activation layer, a batch normalization layer, a deconvolution layer, an activation layer, and a batch normalization layer arranged in sequence; The output of the generative network is an embedding change probability map; The discriminator network model pre-established in the image discrimination module is a combination of steganalysis network models composed of SiaStegNet and SRNet. The neighboring embedded pixel enhancement module includes a weight unit and a collaborative value matrix generation process. Among them, the image discrimination module weight unit It is expressed as: in For weight units at index the value at the index; The process of generating the collaboration value matrix is ​​expressed as: in, The collaboration value matrix is The value of the index, To embed the perturbation graph in The value of the index, u is the horizontal control parameter of the weight unit, and v is the vertical control parameter of the weight unit; The texture consistency loss function of the texture loss module is as follows: in, , are the texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator, respectively. , are the pixel-level texture consistency loss functions corresponding to the positive embedding probability generator and the negative embedding probability generator, respectively. is the image-level texture consistency loss function, is the weight hyperparameter; The pixel-level texture consistency loss function is expressed as: and Represents the carrier image and encrypted images High-frequency features extracted by 30 high-pass filter kernels, Represents pixel-level texture differences, represents the kth filter kernel among the 30 high-pass filter kernels, is an indicator function, indicating The part greater than 0, , , Respectively represent the collaboration value matrix, the positive embedding probability, and the negative embedding probability in The value of the index, represents the gradient calculated by the steganalyzer group. The calculation process is: in, Represents the binary cross entropy loss corresponding to the most discriminative steganalyzer at present, and the input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differential operation; The image-level texture consistency loss function is expressed as: in, , Represent the image-level texture features of the carrier image and the encrypted image, This means that the feature is in the index The value at Express Find the L1-norm, Represents the carrier image Texture features In the index The value at Represents the carrier image Texture features In the index The value at Represents the index of the channel; The input of pixel-level texture loss is the carrier image and encrypted images , the output is pixel-level texture loss; Adaptive binary cross entropy loss function for image steganography module as follows: The input is and , represent the carrier image and the simulated embedded image respectively, Indicates The discriminator takes the input The judgment result of and Represent the true values ​​of the cover image and the simulated encrypted image, It means the differential operation; The collaborative adversarial loss function is as follows: in, is an indicator function, indicating The part greater than 0, , , Respectively represent the collaboration value matrix, the positive embedding probability, and the negative embedding probability in The value of the index, represents the gradient calculated by the steganalyzer group, and Represent the height and length of the image respectively; The loss function of the generator group is as follows: in, , , To control the parameters of each part’s weight, , Represent the total loss function of the positive and negative embedding probability generators, respectively. , Represent the texture consistency loss functions corresponding to the positive and negative embedding probability generators, respectively. represents the pre-set entropy constraint loss function, which is calculated as follows: in Indicates the modification direction. Indicates the embedding rate.

Citation Information

Patent Citations

  • JPEG domain image steganography method and system based on generative adversarial network

    CN110334805A

  • Adversarial network image steganography method based on novel U-Net structure generator

    CN113076549A

  • Image steganography method and system for generating probability model based on U-Net

    CN118279121A

  • Image steganography method and apparatus, and image extraction method and apparatus, and electronic device

    WO2021047471A1