High-safety batch image steganography method

Abstract

The invention relates to a high-security batch image steganography method, which comprises the following steps of: randomly generating n chromosomes, sequencing input images according to the n chromosomes, and generating n groups of images; combining each group of images into a large image, namely a carrier image, and embedding secret information into the large image to obtain n pairs of carrier images and secret images; calculating the similarity between each pair of carrier image and secret image, and selecting the chromosome with the highest similarity as an optimal individual; cloning an optimal individual for t times, selecting n-t individuals from the population according to a similarity measurement value of each pair of carrier images and secret images, performing crossover and mutation operation to obtain n-t optimized individuals, and finally combining t optimal clones with the n-t optimized individuals to form a next-generation population; and repeating selection, crossover and mutation operations, recording the optimal individual of each generation, and taking the chromosome with the highest similarity measurement value as the output of an image stitching scheme ISS.

Description

Technical Field

[0001] The invention belongs to the field of secret communication, and in particular relates to a high-security batch image steganography method. Background Art

[0002] Traditional steganography techniques mostly focus on the steganography method of a single image. Although this method can achieve covert communication to a certain extent, with the advancement of steganalysis technology, especially the widespread application of steganalysis methods based on deep learning, these traditional methods face huge challenges. In addition, among the current mainstream batch image steganography technologies with the highest security, the ES-DD (Embedding Strategy Based on Distortion Distribution) method performs batch steganography based on additive distortion, but does not consider the impact of non-additive distortion on the security of steganography. IMS (Image Merging Sender) uses a single-image steganography algorithm to embed information after splicing images. However, an arbitrary splicing method is used when splicing images, which fails to achieve the optimal solution for batch image steganography security and reduces the security of image steganography. Summary of the invention

[0003] In view of the defects existing in the prior art, the purpose of the present invention is to provide a high-security batch image steganography method, which uses the IMS method to optimize the image splicing scheme when embedding secret information in batch images, and obtains the image splicing scheme with the highest security through the evolution and iteration of the genetic algorithm, so as to approach the optimal solution of the security of batch image steganography as much as possible, thereby improving the security of batch image steganography.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is: a high-security batch image steganography method, the method comprising the following steps:

[0005] S1. Randomly generate n chromosomes in the form of permutation codes, each chromosome represents the order of image merging, and the n chromosomes are regarded as the initial population of the genetic algorithm;

[0006] S2, reorder the input images according to the n chromosomes and generate n groups of images; merge each group of images into a large image, mark each large image as a carrier image, and embed secret information therein to obtain n pairs of carrier images and secret images;

[0007] S3, calculate the similarity of each pair of carrier image and hidden image, and select the chromosome with the highest similarity as the optimal individual;

[0008] S4, clone the best individual t times, and select nt individuals from the population according to the similarity measurement value of each pair of carrier image and secret image, and then perform crossover and mutation operations to obtain nt optimized individuals, and finally combine the t best clones with the nt optimized individuals to form the next generation population;

[0009] S5. Repeat steps S2-S4 for G times in total, and record the best individual of each generation. Select the chromosome with the highest similarity measurement value as the output of the image splicing scheme ISS, and perform image steganography based on the output chromosome.

[0010] Furthermore, in step S2, an image merging module IMS is used to merge each group of images into a large image.

[0011] Furthermore, in step S3, the peak signal-to-noise ratio (PSNR) is used to calculate the similarity of each pair of the carrier image and the stegosaurus image, and the chromosome with the highest peak signal-to-noise ratio (PSNR) is selected as the optimal individual.

[0012] Furthermore, the genetic algorithm in step S4 uses random universal sampling SUS as a selection operator.

[0013] Furthermore, the genetic algorithm in step S4 uses the partial matching crossover algorithm PMX as a crossover operator.

[0014] Furthermore, the genetic algorithm in step S4 uses the two-point exchange algorithm TPS as a mutation operator.

[0015] Further, in step S4, nt individuals are selected from the population according to the peak signal-to-noise ratio (PSNR) of each pair of the carrier image and the stego image.

[0016] Furthermore, step S4 also includes performing a transposition operation on the repeated gene segments of the chromosome after the crossover.

[0017] Further, in step S4, the chromosome crossover probability is set to 100%, and the segment crossover probability is set to 70%.

[0018] Further, in step S5, the chromosome with the highest peak signal-to-noise ratio PSNR is selected as the output of the image stitching scheme ISS.

[0019] The beneficial technical effects of the present invention are: a high-security batch image steganography method disclosed by the present invention, a batch image steganography scheme ISS (Image Stitching Scheme) proposed by the present invention has stronger anti-detection ability than the existing mainstream schemes ES-DD and IMS, retains and further enhances the adaptability of the IMS algorithm and the non-additive embedding method, creatively discovers the influence of the image stitching scheme on the steganography security, and uses a genetic algorithm to realize the optimization of the stitching scheme in the batch image steganography algorithm, realizing the design of a load distribution scheme in multi-image information hiding that is safer and more concealed. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a flow chart of a high-security batch image steganography method shown in the first embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of the process of a PMX crossover method used in a high-security batch image steganography method shown in the first embodiment of the present invention;

[0022] Figure 3 This is a schematic diagram of a TPS mutation operator process used in a high-security batch image steganography method according to the first embodiment of the present invention. DETAILED DESCRIPTION

[0023] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0024] Embodiment 1

[0025] like Figure 1 As shown, an embodiment of the present invention provides a high-security batch image steganography method. ISS (Image Stitching Scheme) optimizes the image stitching scheme using the optimization capability of a genetic algorithm to obtain a multi-image steganography scheme with the strongest anti-detection performance. In the selection of a baseline model, an IMS (Image Merging Sender) model is selected as the basic scheme, and the similarity of the stitched images before and after the secret message is embedded using the IMS model is used as the index function optimized by the genetic algorithm. Appropriate selection operators, crossover operators, and mutation operators are selected to achieve multi-image steganography with higher efficiency and better performance.

[0026] The technical solution principle of the present invention can be expressed by Formula 1:

[0027]

[0028] Among them, Ψ(ISS) represents the ISS load distribution scheme, and Γ(IMS) represents the IMS model in the scheme. If the optimal splicing order Order for the carrier image set X in terms of anti-detection performance can be obtained max , then the above formula in formula (1) can be obtained. In addition, after the secret key information K and the embedding method and extraction method have been determined and transmitted in advance through a secure channel, the steganographic process of the (ISS) scheme can be implemented, that is, the secret information M is embedded into the carrier set X using the ISS method, as shown in the following formula in formula 1.

[0029] like Figure 1 As shown, the method comprises the following steps:

[0030] S1. Randomly generate n permutation coding chromosomes, each chromosome represents the order of image merging, and these n chromosomes are regarded as the initial population of genetic algorithm (GA);

[0031] S2. Reorder the input images according to these chromosomes and generate n groups of images. Use the image merging module (IMS module) to merge each group of images into a large image. Mark each large image as a carrier image and embed secret information in it. In this way, n pairs of carrier images and secret images can be obtained;

[0032] S3. Use the Peak Signal-to-Noise Ratio (PSNR) metric to calculate the similarity of each pair of carrier image and stego image. Select the chromosome with the highest PSNR value as the optimal individual;

[0033] S4. Clone the best individual t times, and select nt individuals from the population according to the PSNR value of each pair of carrier image and secret image. Then, after crossover and mutation operations, nt optimized individuals are obtained. Finally, the t best clones are combined with the nt optimized individuals to form the next generation population.

[0034] S5, repeat steps S2, S3, and S4 for G times, and record the best individual of each generation. Select the chromosome with the highest PSNR value as the output of the image secret sharing (ISS) scheme.

[0035] In the operator selection of the genetic method, SUS (Stochastic Universal Sampling) was selected as the selection operator, PMX (Partially Matched Crossover Algorithm) was selected as the crossover operator, and TPS (Two Point Swapping) was selected as the mutation operator.

[0036] Among them, the SUS selection operator is the best in the experiment compared with other selection operators. In the selection of the crossover operator, since it is a combinatorial optimization problem, each bit of the chromosome must be unique, corresponding to the index value of the carrier image during the merging process. Figure 2 As shown, PMX was chosen as the crossover method. This method is similar to the exchange recombination method, but there is a key difference: when using PMX, it is necessary to ensure that the daughter chromosomes generated after the crossover maintain the characteristics of substitution coding, that is, each bit in the chromosome must be unique. To achieve this, the repeated gene segments of the chromosomes after the crossover are transposed.

[0037] In addition, it has been experimentally verified that the chromosome crossover probability is set to 100% and the segment crossover probability is set to 70%, which means that all chromosomes in the iterative rounds will undergo a crossover operation and 70% of each chromosome will undergo a crossover. It has been experimentally verified that this method can ensure the diversity of the population without losing the genetic advantages of elite individuals.

[0038] like Figure 3 As shown in the figure, in the selection of mutation operators, TPS was selected as the mutation operator due to the particularity of permutation coding. In the experiment, TPS achieved better experimental results than other mutation operators. Similarly, in order to maintain genetic diversity and ensure the genetic advantages of elite individuals, the mutation probability was set to 100% and the probability of mutation fragments was set to 70%, which means that all individuals in the population must undergo mutation operations, and the compiled fragments account for 70% of the total individual length.

[0039] In the selection of fitness function, two common index functions for evaluating image similarity are selected: PSNR (Peak signal-to-noise ratio) and SSIM (Structural Similarity). Experiments have proved that the individuals solved by PSNR have stronger anti-detection ability than SSIM. In addition, in order to explore the effectiveness of the ISS algorithm proposed in the present invention, the entropy of the grayscale co-occurrence matrix of the spliced ​​images before and after being processed by the ISS method is compared in the experiment, which explains in principle why different splicing schemes will affect the security of batch image steganography schemes.

[0040] In addition, in order to explore the optimal batch size and image size, we finally selected an image size of 512×512 and a batch size of 256 images as the steganographic parameters of the ISS batch image steganography scheme. In the experiment, we found that the security of steganography increases with the increase of image size. This is because larger images can give the steganography scheme more flexible load distribution space. Similarly, in the selection of batch size, it is also verified through experiments that the larger the batch size, the stronger the anti-detection performance.

[0041] The batch image steganography scheme proposed in the embodiment of the present invention can show a greater security improvement than the additive embedding algorithm when it is matched with the non-additive embedding algorithm, because the optimization process of the genetic algorithm also has the characteristics of non-additiveness and has a certain ability to escape the local optimal solution. The non-additive embedding algorithm includes HUGO-BD (highly undetectable stego-bounding distortion, a high-performance anti-detection encrypted image generation method based on boundary distortion).

[0042] It can be seen from the above embodiments that a high-security batch image steganography method disclosed by the present invention randomly generates n chromosomes, sorts the input images according to the n chromosomes, and generates n groups of images; merges each group of images into a large image, namely, a carrier image, and embeds secret information therein to obtain n pairs of carrier images and stego images; calculates the similarity of each pair of carrier images and stego images, and selects the chromosome with the highest similarity as the optimal individual; clones the optimal individual t times, and selects nt individuals from the population according to the similarity measurement value of each pair of carrier images and stego images, and then performs crossover and mutation operations to obtain nt optimized individuals, and finally combines the t optimal clones with the nt optimized individuals to form the next generation population; repeats the selection, crossover and mutation operations, and records the optimal individual of each generation, and uses the chromosome with the highest similarity measurement value as the output of the image splicing scheme ISS.

[0043] The method described in the present invention is not limited to the embodiments described in the specific implementation manner. Those skilled in the art may derive other implementation manners based on the technical solution of the present invention, which also fall within the technical innovation scope of the present invention.

Claims

1. A high-security batch image steganography method, the method comprising the following steps: S1. Randomly generate n chromosomes in the form of permutation codes, each chromosome represents the order of image merging, and the n chromosomes are regarded as the initial population of the genetic algorithm; S2, reorder the input images according to the n chromosomes and generate n groups of images; merge each group of images into a large image, mark each large image as a carrier image, and embed secret information therein to obtain n pairs of carrier images and secret images; S3, calculate the similarity of each pair of carrier image and hidden image, and select the chromosome with the highest similarity as the optimal individual; S4, clone the best individual t times, and select nt individuals from the population according to the similarity measurement value of each pair of carrier image and secret image, and then perform crossover and mutation operations to obtain nt optimized individuals, and finally combine the t best clones with the nt optimized individuals to form the next generation population; S5. Repeat steps S2-S4 for G times in total, and record the best individual of each generation. Select the chromosome with the highest similarity measurement value as the output of the image splicing scheme ISS, and perform image steganography based on the output chromosome.

2. A high-security batch image steganography method as claimed in claim 1, characterized in that: In step S2, an image merging module IMS is used to merge each group of images into a large image.

3. A high-security batch image steganography method as claimed in claim 1, characterized in that: In step S3, the peak signal-to-noise ratio (PSNR) is used to calculate the similarity of each pair of carrier image and stego image, and the chromosome with the highest peak signal-to-noise ratio (PSNR) is selected as the optimal individual.

4. A high-security batch image steganography method as claimed in claim 1, characterized in that: The genetic algorithm in step S4 uses random universal sampling SUS as the selection operator.

5. A high-security batch image steganography method as claimed in claim 1, characterized in that: The genetic algorithm in step S4 uses the partial matching crossover algorithm PMX as a crossover operator.

6. A high-security batch image steganography method as claimed in claim 1, characterized in that: The genetic algorithm in step S4 uses the two-point exchange algorithm TPS as a mutation operator.

7. A high-security batch image steganography method as claimed in claim 3, characterized in that: In step S4, nt individuals are selected from the population according to the peak signal-to-noise ratio (PSNR) of each pair of the carrier image and the stego image.

8. A high-security batch image steganography method as claimed in claim 1, characterized in that: Step S4 also includes performing a transposition operation on the repeated gene segments of the chromosome after the crossover.

9. A high-security batch image steganography method as claimed in claim 1, characterized in that: In step S4, the chromosome crossover probability is set to 100%, and the segment crossover probability is set to 70%.

10. A high-security batch image steganography method as claimed in claim 6, characterized in that: In step S5, the chromosome with the highest peak signal-to-noise ratio PSNR is selected as the output of the image stitching scheme ISS.

Images