A visual security image encryption method and device based on parallel compressive sensing and adaptive embedding

Abstract

The application discloses a visual security image encryption method and device based on parallel compressive sensing and adaptive embedding, which comprises the following steps: sparse transformation is performed on a to-be-encrypted plaintext image to obtain a sparse matrix; a chaotic initial value is calculated according to an initial key and sparse matrix coefficients; a chaotic sequence is generated by using Logistics chaotic iteration; a measurement matrix is generated based on the chaotic sequence, and the sparse matrix is scrambled; parallel compressive sensing is performed on each column of the scrambled sparse matrix by using the measurement matrix to obtain a measurement value matrix; the measurement value matrix is quantized and diffused to obtain an intermediate secret image; and the intermediate secret image is adaptively embedded into a carrier image according to regional energy to obtain an embedded carrier image as a ciphertext image. The application uses parallel compressive sensing for pre-encryption, greatly improves the encryption speed, and adaptively embeds the secret image into the carrier image according to the regional energy to obtain the ciphertext image.

Description

Technical Field

[0001] This invention belongs to the field of image encryption technology, specifically relating to a visual security image encryption method and apparatus based on parallel compressed sensing and adaptive embedding. Background Technology

[0002] With the development of technology and the popularization of the Internet, the amount of information on the Internet has exploded. Digital images are widely used in various aspects such as military, politics, and life. The security issues in the transmission of image information have become extremely important. Image encryption is a fast, effective, and widely used technical solution.

[0003] Compressed sensing, proposed by Donoho et al., is a signal sampling theory different from the Nyquist sampling theorem. It enables high-level undersampling, compressing data simultaneously during the sampling process. Image data is characterized by its large volume and high redundancy. Using compressed sensing for image encryption ensures security while compressing the image and reducing transmission bandwidth, making compressed sensing-based image encryption one of the most popular encryption methods today. Since images are two-dimensional data, converting them to one-dimensional signals for processing significantly increases computational complexity. Fang et al. proposed parallel compressed sensing, using a single measurement matrix to sample each column in parallel, greatly improving sampling speed. However, because compressed sensing is a linear projection process, encryption using parallel compressed sensing is vulnerable to plaintext attacks, and the repetitive measurement matrix also reduces security. Hu et al. combined parallel compressed sensing with a counter mode, generating a new measurement matrix for each column. However, the counter mode limits application scenarios, and the measurement matrix construction method used to construct a large number of measurement matrices is very time-consuming. While ensuring security, this approach results in low encryption efficiency. Furthermore, the encryption results of the aforementioned methods are all irregular noise images, which are more likely to attract the attention of hackers when transmitted through insecure channels. To confuse attackers, Bao et al. proposed visual security image encryption, which embeds the encrypted noisy image into a common carrier image to obtain a visually meaningful ciphertext image that is similar to the carrier image. However, the carrier image is usually larger than the plaintext image. Chai applied compressed sensing to visual security encryption, which can also achieve embedding when the carrier image and the plaintext image have the same size. However, the embedding method is relatively fixed, and the visual quality of the resulting ciphertext image is not high. Summary of the Invention

[0004] The purpose of this invention is to provide a visual security image encryption method and apparatus based on parallel compressed sensing and adaptive embedding. Parallel compressed sensing is used for pre-encryption, which greatly improves the encryption speed. The secret image is adaptively embedded into the carrier image according to the regional energy to obtain the ciphertext image.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] This invention provides a visually secure image encryption method based on parallel compressed sensing and adaptive embedding, comprising:

[0007] A sparse matrix is ​​obtained by performing a sparse transformation on the plaintext image to be encrypted.

[0008] Obtain the initialization key, and calculate the initial chaotic value based on the initialization key and the sum of the coefficients of the sparse matrix;

[0009] Based on the aforementioned initial chaotic values, a chaotic sequence is generated using Logistic chaotic iteration;

[0010] A measurement matrix is ​​generated based on the chaotic sequence, and the sparse matrix is ​​scrambled based on the chaotic sequence.

[0011] For each column of the scrambled sparse matrix, parallel compressed sensing is performed using the measurement matrix to obtain the measurement value matrix;

[0012] The intermediate secret image is obtained by quantizing and diffusing the measurement value matrix.

[0013] The intermediate secret image is adaptively embedded into the carrier image based on the regional energy to obtain the embedded carrier image, which serves as the ciphertext image.

[0014] Furthermore, based on the initialization key and the sum of the sparse matrix coefficients, the initial values ​​of chaos are calculated, including:

[0015]

[0016]

[0017] Among them, cx i and cu i Let x represent the initial value of chaos. i and u i This represents the initial key value, sumX. j j = (i-1)×2 + 1, where i×2 represents the first 1-8 decimal places of the sum of sparse coefficients, and sumU j ,j=(i-1)×2+1,i×2 represents the sparse coefficients and the sequence of the 9th to 16th digits after the decimal point.

[0018] Furthermore, the chaotic keys x1, x2, x3, x4 are randomly selected in the range (0, 1), and u1, u2, u3, u4 are randomly selected in the range (3.57, 4).

[0019] Furthermore, based on the aforementioned chaotic initial value, a chaotic sequence is generated using Logistic chaotic iteration, including:

[0020] Based on the aforementioned initial chaotic values, four chaotic sequences seq1, seq2, seq3, and seq4 are generated using Logistic chaotic iteration.

[0021] Where seq1 has a length of row×col, seq2 has a length of (d+1)×col, seq3 has a length of row×col, seq4 has a length of CR×row×col, d is the set interval distance, CR is the set compression ratio, and row×col is the size of the plaintext image to be encrypted.

[0022] Furthermore, generating a measurement matrix based on the chaotic sequence includes:

[0023] The chaotic sequence seq2 is sampled at intervals of distance d. The sampled sequence is used as the initial sequence to construct the first row of the chaotic cycle matrix. The initial sequence is circularly shifted one position to the right to construct the second row of the chaotic cycle matrix. The initial sequence is circularly shifted two positions to the right to construct the third row of the chaotic cycle matrix, and so on, until the row is constructed, to obtain the row×col chaotic cycle matrix matrix.

[0024] Divide the chaotic sequence seq3 into 'col' groups, and sort each group in ascending order to obtain the index sequence for each 'col' group. n where n∈[1,col];

[0025] Using the col group index sequence, partial vectors are selected from the chaotic cyclic matrix matrix to form part of the chaotic cyclic matrix as measurement matrices, generating a total of col measurement matrices, as shown below:

[0026] Phi n (i,:) = matrix(index) n (i),:);

[0027] Among them, Phi n For the nth measurement matrix, index n (i) represents the i-th value of the n-th index sequence, i = 1, 2, ..., CR×row.

[0028] Furthermore, scrambling the sparse matrix based on the chaotic sequence includes:

[0029] The chaotic sequence seq1 is sorted in ascending order using the sort function to obtain the sort index index_P;

[0030] Transform the sparse matrix into a one-dimensional sparse vector P1;

[0031] The one-dimensional sparse vector P1 is scrambled using the sorting index index_P, and the scrambled vector is then converted into a two-dimensional sparse matrix P2.

[0032] Furthermore, for each column of the scrambled sparse matrix, parallel compressed sensing is performed using the measurement matrix to obtain a measurement value matrix, including:

[0033] Using col unique measurement matrices, compressed sensing is performed on the corresponding columns of the two-dimensional sparse matrix P2 in parallel to obtain the measurement value matrix P3.

[0034] Furthermore, the measured value matrix is ​​quantized and diffused to obtain an intermediate secret image, including:

[0035] The measured value matrix is ​​quantized, and the pixel values ​​are quantized to the range (0, 255), represented as follows:

[0036] C(i,j)=round(255×(P3(i,j)-min) / (max-min));

[0037] Where C(i,j) is the quantized value at position (i,j) in the measurement matrix P3, P3(i,j) is the pixel value at position (i,j) in the measurement matrix P3, min is the minimum pixel value in the measurement matrix P3, and max is the maximum pixel value in the measurement matrix P3.

[0038] The chaotic sequence seq4 is converted into a (CR×row)×col chaotic matrix. The quantized measurement matrix is ​​then diffused using this chaotic matrix, once in each direction, to obtain the intermediate secret image C3, as shown below:

[0039]

[0040] Where C3(i,j) represents the pixel value at position (i,j) in the intermediate secret image C3, C2(i,j) represents the pixel value at position (i,j) in the image after forward diffusion of the quantized measurement matrix, and seq4(i,j) represents the value at position (i,j) in the chaotic matrix. This indicates the XOR operation.

[0041] Furthermore, the step of adaptively embedding the intermediate secret image into the carrier image based on the region energy to obtain the embedded carrier image as the ciphertext image includes:

[0042] The carrier image is subjected to integer wavelet transform, and the high-frequency sparse subbands LH, HL and HH of the transformed matrix are divided into blocks to obtain carrier image sub-blocks.

[0043] Calculate the region energy of each carrier image sub-block and sort the carrier image sub-blocks in descending order of region energy;

[0044] The intermediate secret image is divided into blocks, and the CR value is set so that the number of secret image sub-blocks is one-third of the number of carrier image sub-blocks;

[0045] Embedding is performed based on the regional energy of the carrier image sub-blocks. The units digit of the secret image sub-block is embedded in one-third of the carrier image sub-blocks with the largest regional energy, the hundreds digit of the secret image sub-block is embedded in one-third of the carrier image sub-blocks with the smallest regional energy, and the tens digit of the secret image sub-block is embedded in the remaining carrier image sub-blocks.

[0046] The embedded carrier image sub-blocks are merged and subjected to inverse integer wavelet transform to obtain the final encrypted image.

[0047] The present invention also provides a visual security image encryption device based on parallel compressed sensing and adaptive embedding, for implementing the aforementioned visual security image encryption method based on parallel compressed sensing and adaptive embedding, the device comprising:

[0048] The transformation module is used to perform sparse transformation on the plaintext image to be encrypted to obtain a sparse matrix;

[0049] The initial value acquisition module is used to acquire the initialization key and calculate the initial chaotic value based on the initialization key and the sum of the coefficients of the sparse matrix.

[0050] A chaotic sequence generation module is used to generate a chaotic sequence based on the initial chaotic value using Logistic chaotic iteration;

[0051] A measurement matrix generation module is used to generate a measurement matrix based on the chaotic sequence;

[0052] A scrambling module is used to scramble the sparse matrix based on the chaotic sequence;

[0053] The compressed sensing module is used to perform parallel compressed sensing on each column of the scrambled sparse matrix using the measurement matrix to obtain a measurement value matrix.

[0054] The quantization and diffusion module is used to quantize and diffuse the measurement value matrix to obtain an intermediate secret image;

[0055] An embedding module is used to adaptively embed the intermediate secret image into the carrier image based on the regional energy, so as to obtain the embedded carrier image as the ciphertext image.

[0056] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0057] 1. This invention uses a partially chaotic cyclic matrix to construct the measurement matrix, which can quickly construct the required measurement matrix, improve encryption efficiency, and solve the problem that the measurement matrix is ​​difficult to transmit and store as a key.

[0058] 2. In the parallel compressed sensing process, the present invention uses a different measurement matrix when measuring each column, which better ensures the security of encryption compared to using a fixed measurement matrix for parallel compressed sensing encryption.

[0059] 3. This invention uses a fixed key and the sparse coefficients of the image and updates the initial value of the chaos, which can effectively resist plaintext attacks;

[0060] 4. This invention adaptively embeds the secret image into the carrier image based on the regional energy, resulting in a ciphertext image with better visual quality. Attached Figure Description

[0061] Figure 1 A flowchart of a visual security image encryption method based on parallel compressed sensing and adaptive embedding is provided in an embodiment of the present invention.

[0062] Figure 2 This is an example of constructing a chaotic cycle matrix and generating a measurement matrix in an embodiment of the present invention. Detailed Implementation

[0063] The present invention will now be further described. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0064] This invention provides a visually secure image encryption method based on parallel compressed sensing and adaptive embedding, see [link to relevant documentation]. Figure 1 ,include:

[0065] A sparse matrix is ​​obtained by performing a sparse transformation on the plaintext image to be encrypted.

[0066] Initialize the key and calculate the initial values ​​of chaos based on the key and the sum of the coefficients of the sparse matrix;

[0067] Based on the calculated initial chaotic values, a chaotic sequence is generated using Logistic chaotic iteration;

[0068] A measurement matrix is ​​generated based on the generated chaotic sequence;

[0069] The sparse matrix is ​​scrambled based on the generated chaotic sequence;

[0070] For each column of the scrambled sparse matrix, parallel compressed sensing is performed using the measurement matrix to obtain the measurement value matrix;

[0071] The intermediate secret image is obtained by quantizing and diffusing the measurement matrix;

[0072] The intermediate secret image is adaptively embedded into the carrier image based on the regional energy, resulting in the embedded carrier image, which serves as the ciphertext image.

[0073] In one embodiment of the present invention, a sparse matrix is ​​obtained by performing a sparse transformation on a plaintext image. The specific implementation process is as follows:

[0074] Assuming the plaintext image P has dimensions row×col, a sparse transformation is performed on the plaintext image P to obtain a sparse matrix P1. The transformation method is defined as follows:

[0075] P1 = Ψ × P × Ψ′;

[0076] Where P represents the plaintext image, Ψ represents the sparse basis, Ψ′ represents the transpose of Ψ, and P1 represents the sparse matrix.

[0077] In one embodiment of the present invention, the key is initialized and the initial value of chaos is calculated based on the key and the sum of the coefficients of the sparse matrix. The specific implementation process is as follows:

[0078] The initialization keys include: the interval distance d, the compression ratio CR, and the chaotic keys x1, x2, x3, x4 and u1, u2, u3, u4;

[0079] The initial values ​​of chaos are calculated using the chaos key and the coefficients of the sparse matrix, as follows:

[0080]

[0081]

[0082] Among them, cx i and cu i Let sumX represent the initial value of chaos. j j = (i-1)×2 + 1, where i×2 represents the first 1-8 decimal places of the sum of sparse coefficients, and sumU j ,j=(i-1)×2+1,i×2 represents the sparse coefficients and the sequence of the 9th to 16th digits after the decimal point.

[0083] It should be noted that the chaotic keys x1, x2, x3, x4 are randomly selected in the range (0, 1), and u1, u2, u3, u4 are randomly selected in the range (3.57, 4).

[0084] In one embodiment of the present invention, a chaotic sequence is generated using Logistic chaotic iteration based on the calculated initial chaotic value, as follows:

[0085] Based on the calculated initial chaotic values, four chaotic sequences seq1, seq2, seq3, and seq4 are generated using Logistic chaotic iteration.

[0086] Logistics chaos is defined as:

[0087] x n+1 =u·x n (1-x n );

[0088] Where u is a control parameter, and its value ranges from 3.57. <u<4。

[0089] Among them, seq1 with length row×col is used for scrambling; seq2 and seq3 with length (d+1)×col are used to generate a partial chaotic cycle matrix; seq4 with length CR×row×col is used for diffusion; d is the set interval distance, and CR is the set compression ratio.

[0090] In one embodiment of the present invention, a measurement matrix is ​​generated based on the generated chaotic sequence, and the specific implementation process is as follows:

[0091] S11. Sample the chaotic sequence seq2 at intervals d, and use the sampled sequence as the initial sequence to construct the chaotic cycle matrix matrix.

[0092] The specific process of constructing a chaotic cyclic matrix is ​​as follows: Let the sampled sequence be S. Construct a chaotic cyclic matrix. The first row is S, the second row is S shifted one position to the right, the third row is S shifted two positions to the right, and so on, until the row is reached. This constructs a row × col matrix. See also... Figure 2 This shows the process of constructing a chaotic cycle matrix.

[0093] S12. Divide the chaotic sequence seq3 into 'col' groups of sequences, and sort each group in ascending order to obtain the index sequence of the 'col' groups. n where n∈[1,col], each index n The length of the sequence is row.

[0094] S13. Using the col group index sequence, select a portion of the vectors from the chaotic cyclic matrix matrix to form the chaotic cyclic matrix as the measurement matrix, generating a total of col measurement matrices. The selection method is as follows:

[0095] Phi n (i,:) = matrix(index) n (i),:);

[0096] Among them, Phi n For the nth measurement matrix, index n Let n be the index sequence of the nth group, where n = 1, 2, ..., col, and index n (i) represents the i-th value of the n-th index sequence, i = 1, 2, ..., CR×row.

[0097] See Figure 2 This shows the process of selecting a subset of vectors to form a measurement matrix.

[0098] In one embodiment of the present invention, the sparse matrix is ​​scrambled based on the generated chaotic sequence, specifically as follows:

[0099] Since the energy distribution of the sparse matrix after sparse transformation is relatively concentrated, a scrambling operation is needed to make the energy distributed randomly.

[0100] S21. Sort the chaotic sequence seq1 in ascending order to obtain the sort index index_P. In this embodiment, the sort function is used to sort the chaotic sequence seq1, as shown below:

[0101] [~,index_P] = sort(seq1);

[0102] S22. Transform the sparse matrix into a one-dimensional sparse vector, as follows:

[0103] P1 = P1(:);

[0104] S23. Use the sorting index index_P to scramble the sparse vector P1, and then convert the scrambled vector into a two-dimensional sparse matrix P2, as shown below:

[0105]

[0106] Where P2(i) represents the i-th element in the scrambled one-dimensional vector, P1(index_P(i)) represents the index_P(i)-th element in the one-dimensional sparse vector P1, index_P(i) represents the i-th element in the sorted index index_P, and P2 represents the scrambled two-dimensional sparse matrix.

[0107] In one embodiment of the present invention, for each column of the scrambled sparse matrix, parallel compressed sensing is performed using a measurement matrix to obtain a measurement value matrix. The specific implementation process is as follows:

[0108] Use col unique measurement matrices {Phi1, Phi2, ..., Phi col In parallel, compressed sensing is performed on the corresponding columns of the two-dimensional sparse matrix P2. The compressed sensing process is represented as follows:

[0109] P3(:,n)=Phi n ×P2(:,n);

[0110] Where P3 is the measurement matrix after compressed sensing, and P2(:,n) represents the nth column of matrix P2, where n = 1, 2, ..., col.

[0111] In one embodiment of the present invention, the measurement value matrix is ​​quantized and diffused to obtain an intermediate secret image, and the specific operation is as follows:

[0112] In this embodiment, the measurement value matrix needs to be quantized to the range of (0, 255). In order to achieve uniform pixel distribution, the matrix is ​​finally subjected to bidirectional diffusion. The specific implementation steps are as follows:

[0113] S31. Quantize the measurement value matrix, quantizing the pixel values ​​to the range (0, 255), to obtain the quantized measurement value matrix C, represented as:

[0114] C(i,j)=round(255×(P3(i,j)-min) / (max-min));

[0115] Where C(i,j) is the quantized value at position (i,j) in the measurement matrix P3, P3(i,j) is the pixel value at position (i,j) in the measurement matrix P3, min is the minimum pixel value in the measurement matrix P3, and max is the maximum pixel value in the measurement matrix P3.

[0116] S32. Convert the chaotic sequence seq4 into a chaotic matrix of (CR×row)×col. Apply the chaotic matrix diffusion to the quantized measurement matrix C, performing it once in each direction to obtain the intermediate secret image C3, as shown below:

[0117]

[0118] Where C3(i,j) represents the pixel value at position (i,j) in the intermediate secret image, C2 is the image after forward diffusion of the measurement matrix C, C2(i,j) represents the pixel value at position (i,j) in the forward-diffused image C2, and seq4(i,j) represents the value at position (i,j) in the chaotic matrix. This indicates the XOR operation.

[0119] In one embodiment of the present invention, the intermediate secret image is adaptively embedded into the carrier image based on the region energy to obtain the embedded carrier image, which serves as the ciphertext image. See [link to previous document]. Figure 1 Specifically, it includes:

[0120] S41. Perform integer wavelet transform on the carrier image, and divide the high-frequency sparse subbands LH, HL and HH of the transformed matrix into blocks, each block being 4×4, for a total of 3×row×col / 64 carrier image sub-blocks.

[0121] S42. Calculate the region energy of each carrier image sub-block:

[0122]

[0123] Where F(x,y) represents the pixel value at position (x,y) of the carrier image sub-block, and m×n represents the size of the carrier image sub-block;

[0124] S43. Obtain the sorting sequence of region energy by dividing the energy of the carrier image sub-blocks from largest to smallest;

[0125] S44. Divide the middle secret image into blocks, each block being 4×4 in size. Let CR be 1 / 4, then you get row×col / 64 secret image sub-blocks.

[0126] S45. Embedding is performed based on the region energy of the carrier image sub-blocks. There are a total of 3 × row × col / 64 carrier image sub-blocks. The units digit of the secret image sub-block is embedded in the row × col / 64 carrier image sub-blocks with the largest region energy. The hundreds digit of the secret image sub-block is embedded in the row × col / 64 carrier image sub-blocks with the smallest region energy. The tens digit of the secret image sub-block is embedded in the remaining row × col / 64 carrier image sub-blocks. The embedding formula is:

[0127] emb(i,j)=car(i,j)-(car(i,j)mod10)+U(i,j)

[0128] emb(i,j)=car(i,j)-(car(i,j)mod10)+T(i,j);

[0129] emb(i,j)=car(i,j)-(car(i,j)mod10)+H(i,j)

[0130] Where, emb(i,j) is the pixel at position (i,j) of the embedded secret image sub-block, car(i,j) is the pixel at position (i,j) of the carrier image sub-block, U(i,j) is the units digit of the pixel value at position (i,j) of the secret image sub-block, T(i,j) is the tens digit of the pixel value at position (i,j) of the secret image sub-block, and H(i,j) is the hundreds digit of the pixel value at position (i,j) of the secret image sub-block;

[0131] S46. Merge the embedded carrier image sub-blocks and perform integer wavelet inverse transform to obtain the final ciphertext image.

[0132] Based on the same inventive concept, the present invention also provides a visual security image encryption device based on parallel compressed sensing and adaptive embedding, for implementing the visual security image encryption method based on parallel compressed sensing and adaptive embedding of the foregoing embodiments. The device includes:

[0133] The transformation module is used to perform sparse transformation on the plaintext image to be encrypted to obtain a sparse matrix;

[0134] The initial value acquisition module is used to acquire the initialization key and calculate the initial chaotic value based on the initialization key and the sum of the coefficients of the sparse matrix.

[0135] A chaotic sequence generation module is used to generate a chaotic sequence based on the initial chaotic value using Logistic chaotic iteration;

[0136] A measurement matrix generation module is used to generate a measurement matrix based on the chaotic sequence;

[0137] A scrambling module is used to scramble the sparse matrix based on the chaotic sequence;

[0138] The compressed sensing module is used to perform parallel compressed sensing on each column of the scrambled sparse matrix using the measurement matrix to obtain a measurement value matrix.

[0139] The quantization and diffusion module is used to quantize and diffuse the measurement value matrix to obtain an intermediate secret image;

[0140] An embedding module is used to adaptively embed the intermediate secret image into the carrier image based on the regional energy, so as to obtain the embedded carrier image as the ciphertext image.

[0141] It is worth noting that this device embodiment corresponds to the above method embodiment. The implementation methods of the above method embodiments are all applicable to this device embodiment and can achieve the same or similar technical effects, so they will not be described in detail here.

[0142] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0143] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0144] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0145] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A visually secure image encryption method based on parallel compressed sensing and adaptive embedding, characterized in that, include: A sparse matrix is ​​obtained by performing a sparse transformation on the plaintext image to be encrypted. Obtain the initialization key, and calculate the initial value of chaos based on the initialization key and the sum of coefficients of the sparse matrix; Based on the initial chaotic value, a chaotic sequence is generated using Logistic chaotic iteration, represented as: seq1, seq2, seq3, seq4; Generating a measurement matrix based on the chaotic sequence includes: The chaotic sequence seq2 is sampled at intervals of distance d. The sampled sequences are used as the initial sequence to construct the first row of a chaotic cycle matrix. The initial sequence is then circularly shifted one position to the right to construct the second row of the chaotic cycle matrix. The initial sequence is then circularly shifted two positions to the right to construct the third row of the chaotic cycle matrix, and so on, until the... Okay, constructing × The chaotic cyclic matrix matrix; where, × The size of the plaintext image to be encrypted; Divide the chaotic sequence seq3 into equal parts Group sequences, sort each group in ascending order to obtain Group Index Sequence ,in ; use The group index sequence selects a portion of the vectors from the chaotic cyclic matrix matrix to form the component chaotic cyclic matrix as the measurement matrix, generating a total of A measurement matrix, represented as follows: ； in, For the first A measurement matrix, Representing the The first group index sequence One value, =1,2,…,CR× ;CR is the compression ratio set; The sparse matrix is ​​scrambled based on the chaotic sequence; For each column of the scrambled sparse matrix, parallel compressed sensing is performed using the measurement matrix to obtain the measurement value matrix; The intermediate secret image is obtained by quantizing and diffusing the measurement value matrix. The intermediate secret image is adaptively embedded into the carrier image based on the region energy to obtain the embedded carrier image, which serves as the ciphertext image, including: The carrier image is subjected to integer wavelet transform, and the high-frequency sparse subbands LH, HL and HH of the transformed matrix are divided into blocks to obtain carrier image sub-blocks. Calculate the region energy of each carrier image sub-block and sort the carrier image sub-blocks in descending order of region energy; The intermediate secret image is divided into blocks, and the CR value is set so that the number of secret image sub-blocks is one-third of the number of carrier image sub-blocks; Embedding is performed based on the region energy of the carrier image sub-blocks. The units digit of the secret image sub-block is embedded in the first third of the carrier image sub-blocks with the highest region energy; the hundreds digit of the secret image sub-block is embedded in the first third of the carrier image sub-blocks with the lowest region energy; and the tens digit of the secret image sub-block is embedded in the remaining carrier image sub-blocks. The embedding formula is expressed as: ； in, For the location of the embedded secret image sub-block Pixels at that location, Location of carrier image sub-blocks Pixels at that location, For the location of secret image sub-blocks The units digit of the pixel value. For the location of secret image sub-blocks The ten digits of the pixel value. For secret image location sub-blocks The hundreds digit of the pixel value; The embedded carrier image sub-blocks are merged and subjected to inverse integer wavelet transform to obtain the final encrypted image.

2. The visual security image encryption method based on parallel compressed sensing and adaptive embedding according to claim 1, characterized in that, Based on the initialization key and the sum of the sparse matrix coefficients, calculate the initial values ​​of chaos, including: ， ； ， ； in, and Represents the initial value of chaos. and Indicates the initial value of the key. This represents the first 1-8 digits after the decimal point of the sum of the sparse coefficients. This represents the sparse coefficients and the sequence of digits from the 9th to the 16th decimal place.

3. The visual security image encryption method based on parallel compressed sensing and adaptive embedding according to claim 2, characterized in that, Chaos Key Randomly select within the range (0,1). Randomly select from the range (3.57,4).

4. The visual security image encryption method based on parallel compressed sensing and adaptive embedding according to claim 3, characterized in that, Based on the aforementioned chaotic initial value, a chaotic sequence is generated using Logistic chaotic iteration, including: Based on the aforementioned initial chaotic values, four chaotic sequences seq1, seq2, seq3, and seq4 are generated using Logistic chaotic iteration. Where, the length of seq1 is × The length of seq2 is (d+1)× The length of seq3 is × The seq4 length is CR× × d is the set interval distance, and CR is the set compression ratio. × The size of the plaintext image to be encrypted.

5. A visual security image encryption method based on parallel compressed sensing and adaptive embedding according to claim 4, characterized in that, The scrambling of the sparse matrix based on the chaotic sequence includes: The chaotic sequence seq1 is sorted in ascending order using the sort function to obtain the sort index. ; Transform the sparse matrix into a one-dimensional sparse vector. ; Use sorted index For the one-dimensional sparse vector Scramble the vectors and convert the scrambled vectors into two-dimensional sparse matrices. .

6. A visual security image encryption method based on parallel compressed sensing and adaptive embedding according to claim 5, characterized in that, For each column of the scrambled sparse matrix, parallel compressed sensing is performed using the measurement matrix to obtain a measurement value matrix, including: use A unique measurement matrix, and parallel processing of the two-dimensional sparse matrix. Compressed sensing is performed on the corresponding columns to obtain the measurement value matrix. .

7. A visual security image encryption method based on parallel compressed sensing and adaptive embedding according to claim 6, characterized in that, The intermediate secret image is obtained by quantizing and diffusing the measured value matrix, including: The measured value matrix is ​​quantized, and the pixel values ​​are quantized to the range (0, 255), represented as follows: ； in, It is a matrix of measured values. Location The quantized value, It is a matrix of measured values. Location The pixel value at that location, min is the measurement matrix. The minimum pixel value in the matrix, where max is the measurement value matrix. The maximum pixel value in; Convert the chaotic sequence seq4 to (CR×) )× The chaotic matrix is ​​used to diffuse the quantized measurement matrix, once in each direction, to obtain the intermediate secret image. , means as follows: ； in, Representing the intermediate secret image Location Pixel value at that location, The image represents the forward diffusion of the quantized measurement matrix. Location Pixel value at that location, Represents the position of the chaotic matrix The value at that location, This indicates the XOR operation.

8. A visual security image encryption device based on parallel compressed sensing and adaptive embedding, characterized in that, The apparatus for implementing the visual secure image encryption method based on parallel compressed sensing and adaptive embedding as described in any one of claims 1 to 7, the apparatus comprising: The transformation module is used to perform sparse transformation on the plaintext image to be encrypted to obtain a sparse matrix; The initial value acquisition module is used to acquire the initialization key and calculate the initial chaotic value based on the initialization key and the sum of the coefficients of the sparse matrix. A chaotic sequence generation module is used to generate a chaotic sequence based on the initial chaotic value using Logistic chaotic iteration; A measurement matrix generation module is used to generate a measurement matrix based on the chaotic sequence; A scrambling module is used to scramble the sparse matrix based on the chaotic sequence; The compressed sensing module is used to perform parallel compressed sensing on each column of the scrambled sparse matrix using the measurement matrix to obtain a measurement value matrix. The quantization and diffusion module is used to quantize and diffuse the measurement value matrix to obtain an intermediate secret image; An embedding module is used to adaptively embed the intermediate secret image into the carrier image based on the regional energy, so as to obtain the embedded carrier image as the ciphertext image.

Images