Hyperchaotic parallel encryption method for images in network office approval documents

By generating chaotic sequences with anti-chaotic degradation properties through a cascaded multidimensional chaotic system and the RANSAC-circle algorithm, a chaotic cryptography generator is constructed for image encryption, solving the problems of insufficient security and efficiency in existing technologies and achieving efficient and secure image encryption.

CN114465706BActive Publication Date: 2024-11-22HARBIN UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111453862.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-01
Publication Date
2024-11-22
Estimated Expiration
2041-12-01

AI Technical Summary

Technical Problem

Existing chaotic image encryption methods are insufficient in terms of security and efficiency, are difficult to effectively resist malicious attacks, and have poor computational performance.

Method used

A cascaded multidimensional chaotic system and the RANSAC-circle algorithm are used to generate chaotic sequences with anti-chaotic degradation properties through parallel computation. A chaotic cryptography generator is constructed, and three sets of iterative keys are used for image encryption, including 2D-Logistic key, RANSAC-circle key and superLorenz key, for preprocessing, scrambling and diffusion operations.

Benefits of technology

It improves the security and efficiency of image encryption, enhances key sensitivity, effectively resists statistical and differential attacks, ensures the security of images during network transmission, and improves computing speed.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114465706B_ABST
    Figure CN114465706B_ABST
Patent Text Reader

Abstract

The present invention proposes a method for parallel encryption of images in network office approval documents based on hyperchaos. It includes: 1. A chaotic password generator is composed of a key string, and the key string is composed of three groups of mutually iterative keys: 2D-Logistic key, RANSAC-circle key and superLorenz key. 2. Receive network electronic document images, and adopt a pre-processing-diffusion-scrambling-diffusion architecture mode in the encryption process to achieve block parallel encryption. The present invention effectively improves the sensitivity of image keys, reduces the calculation time for encrypting images with large data, expands the key space, improves the calculation speed of the encryption process, and effectively resists hacker attacks such as statistical attacks and differential attacks; the present invention solves the problem of data inconsistency caused by the calculation of different word lengths in network transmission, ensures the security of cross-transmission of sensitive document image data in public networks, and can be applied to information security application fields such as network approval documents.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of image encryption and parallel computing, and in particular relates to a novel parallel digital image encryption method in network office approval documents based on hyperchaotic maps. Background Art

[0002] With the development of distributed computer networks, digital image processing, and mobile storage devices, digital images, as one of the important carriers of electronic information dissemination today, have been widely used in many fields. Due to the cross-transmission of information on public networks, the transmission process is vulnerable to a variety of security threats, such as illegal theft, malicious tampering, and unauthorized copying. Therefore, a large number of potential dangers in the transmission of digital images on public networks have attracted widespread attention from researchers in the field of information security.

[0003] The essence of digital image encryption is to quickly convert the actual image into an image form that cannot be interpreted by the outside world before it is transmitted across the public network. The converted image is actually a collection of image attributes whose image information is obscured by chaos. Therefore, an excellent digital image encryption algorithm must have the characteristics of confidentiality, integrity, verifiability, non-repudiation and high speed.

[0004] The current digital image encryption methods can be summarized into four categories: optical encryption, transform domain encryption, spatial domain encryption, and compressed sensing. The basis of spatial domain encryption methods includes chaotic mapping, DNA, elliptic curves, etc. Chaotic mapping in a dynamic environment can produce a large change in output data after a small change in the initial conditions. Chaotic mapping is divided into continuous mapping and discrete mapping. It has been widely used in data security transmission and communication, and has a good balance between computing speed and security. In the application of image encryption, chaotic mapping roughly divides the image encryption process into two stages: scrambling and diffusion. Scrambling refers to the use of the pseudo-random chaotic sequence generated by chaotic mapping to scramble the plaintext image in units of pixels, solving the key dependency problem in encryption. Diffusion refers to the scrambled image in units of blocks, and the preconditions of the chaotic mapping are updated through the previous blocks in the process.

[0005] An ideal image encryption method can ensure that image data in most application systems involving image data security can better resist different intrusion attacks. Among them, due to the characteristics of chaos-based image encryption, it has attracted the attention of a large number of researchers. Researchers have studied encryption methods from the perspective of time domain, space domain, dimension, parameter and other variables. For example, it is proposed to improve the key space using iterative chaotic mapping of two-dimensional sine mapping and infinite collapse modulation mapping (i.e., 2D-SIMM), and merge scrambling and diffusion into one step to reduce the calculation time. In addition, a chaotic system that is scrambled by using the difference in the output sequences of two identical chaotic systems is proposed, and a scrambling-diffusion-scrambling encryption architecture is adopted. This technology solves the shortcomings of a single chaotic mapping. Similarly, some scholars have improved the encryption system into four encryption stages: diffusion-scrambling-diffusion-permutation. These stages improve the overall encryption performance by enhancing chaotic mapping, S-box, Logistic mapping and permutation algorithms. In the horizontal technology comparison, the encryption speed level is improved, and the use of parallel computing technology improves the calculation speed of chaotic cryptography in the field of image encryption. Its significant encryption performance is favored by many scholars.

[0006] Due to the ergodicity, sensitivity, pseudo-randomness, universality, nonlinearity and long-term unpredictability of chaotic systems, they are completely consistent with the essence of digital image encryption. Therefore, scholars have studied a large number of different chaotic image encryption methods and applied them in various scenarios. Since the key space of low-dimensional chaotic mapping is small and the security is poor, using high-dimensional chaotic mapping as the basis for digital image encryption will greatly improve the security of encrypted images. However, compared with spatial domain encryption technology, although the image encryption method based on hyperchaos meets the performance indicators of image encryption, its computing performance is poor. Therefore, the use of parallel computing methods and the use of GPUs to implement parallel encryption methods for digital graphics based on hyperchaos can increase the encryption computing performance of large data sets. The applications and advantages of parallel digital image encryption based on hyperchaos are listed below.

[0007] Case 1: An office needs to review electronic documents online through the Internet. Since the electronic documents contain internal reports from each office, they need to be transmitted and stored through the Internet after being reviewed. There are security loopholes in this process, which can easily be illegally stolen by outsiders and cause key information leakage.

[0008] Case 2: A hospital needs to protect the medical images of patients. The various departments of the hospital store images of the patient's body and some of its internal tissues for doctors to treat patients and researchers to study diseases. In the process of treatment, research and communication, it needs to be disseminated on the public network. In this process, since medical images are an important basis for patient treatment and research, it is easy to cause security leaks of patient privacy information and research materials. For example, medical images containing bones have many continuous and same-color pixels during the encryption process, and most medical images are large in size, high in resolution, high in pixel accuracy, and require a large amount of data to be encrypted. Therefore, the use of medical image encryption based on parallel hyperchaos is highly secure and fast in calculation, which can solve the above problems.

[0009] Case 3: A research institute performs security protection on the transmission and storage of space remote sensing images. Remote sensing images contain a large amount of important strategic information, such as geographic information, military information, and meteorological information. In the process of using the corresponding earth observation satellite to transmit remote sensing images to the ground and the research institute to store remote sensing images, a series of security issues such as malicious attacks and unauthorized access may occur, resulting in the theft of a large amount of key data and undermining data security.

[0010] Case 4: A cloud storage service company provides security protection for images uploaded by users. At present, public cloud storage platforms provide users with convenient and low-cost storage services, but there are a lot of security risks in the storage of users' multimedia data on cloud platforms. In order to improve the strength of privacy security, a method that is both safe, reliable and fast is needed. The advantages of this method, which is efficient and reliable, can help cloud platforms solve multimedia data security issues.

[0011] In summary, the method of implementing a new type of parallel digital image encryption and protection in network office approval documents using chaotic random sequences based on hyperchaotic mapping can achieve secure transmission of sensitive information. Summary of the invention

[0012] 1. Technical issues to be resolved

[0013] In view of the security problems and efficiency deficiencies of current chaotic image encryption methods, the present invention provides a novel parallel digital image encryption method based on hyperchaos, which improves the security, encryption efficiency, correlation between plaintext and ciphertext, key space, and sensitivity of plaintext and ciphertext of image encryption. The ciphertext image histogram is more uniform, which can effectively resist various malicious attacks.

[0014] In order to overcome the chaotic system degradation problem caused by limited precision, the present invention proposes the following technical solution, which has good anti-chaotic degradation characteristics, as shown in:

[0015] 1. The present invention increases the period length of the chaotic system by cascading two multidimensional chaotic systems and the RANSAC-circle (RANSAC): random sampling consensus algorithm) algorithm, and the present invention respectively generates sequences through periodic sinusoidal perturbations of the two chaotic systems and the associated RANSAC-circle algorithm, so that the two multidimensional chaotic systems have stronger anti-chaotic degradation characteristics and will not overlap with each other.

[0016] 2. In the process of 2D-Logistic key generation sequence, the initial values ​​of the chaotic system are x0, y0, and the step size is h. The initial values ​​are sinusoidally disturbed every 3000 cycles to produce anti-chaotic degradation characteristics.

[0017] 3. In the RANSAC-circle algorithm, since the data set is different each time, the sequence value generated by 2D-Logistic is used as the model radius and noise coefficient, which will not cause degradation and effectively combat the periodic distribution characteristics of the sequence value.

[0018] 4. In the multidimensional Lorenz chaotic system, based on the sequence value generated by the RANSAC-circle algorithm, a sinusoidal disturbance is performed every 3000 cycles. After the disturbance signal is generated, it is superimposed on the original chaotic orbit according to the step size, which increases the orbit length of the multidimensional Lorenz chaotic system and produces strong anti-chaotic degradation characteristics.

[0019] (II) Technical solution

[0020] In order to solve the above technical problems and improve encryption performance, the present invention provides a parallel digital image encryption method based on hyperchaos.

[0021] The basic idea of ​​the parallel digital image encryption method based on hyperchaos is: the plaintext association scrambling algorithm is adopted as a whole to increase the correlation between the key and the plaintext and thus enhance security. At the same time, a "key chain" with anti-chaotic degradation characteristics is formed by a cascaded multidimensional chaotic system and RANSAC. A chaotic image encryption password generator is generated through parallel computing. The corresponding actions in the process obtain the required chaotic sequence from the generator to perform block-by-block parallel encryption on the plaintext image, and finally the corresponding encrypted image is obtained by integration.

[0022] It includes the following specific steps:

[0023] 1. Constructing a Chaotic Code Generator

[0024] The overall encryption process requires a pseudo-random chaotic sequence generated by a chaotic cipher generator for scrambling and diffusion operations in the encryption process. The chaotic cipher generator is constructed from a "key string", which consists of three sets of mutually iterative "keys": 2D-Logistic key, RANSAC-circle key, and superLorenz key. The four sequences generated by the chaotic cipher generator are used for preprocessing, scrambling, diffusion and other operations in the encryption process.

[0025] First, receive the image in the electronic document on the network, symbolize it as P for computer representation, and if the length and width m and n are not divisible, correct them with 0 pixel values. After correction, the image size is m×n. The key generation steps are as follows:

[0026] Func1.1, 2D-Logistic key

[0027] The key is based on the two-dimensional Logistic (i.e. 2D-Logistic) mapping as the initial iteration key of the "key string". The specific steps are described as follows:

[0028] S1_1_1, input the initial parameters u1 and u2 of the two-dimensional Logistic chaotic map;

[0029] S1_1_2, substitute into E1 and iterate m×n+800 times, adding the periodic perturbation of E2 every 3000 iterations;

[0030]

[0031] E2:x0=x0+h×sin(y0)

[0032] Among them, E1 is the two-dimensional Logistic chaos mapping equation, E2 is the periodic sinusoidal perturbation formula; x0 and y0 are the initial state variables of the mapping, x i ,y i is the state variable of the mapping, u1 and u2 are system parameters, h is the mapping step size, and when the system parameter value range is (0,4], the system is in a chaotic state. (X mod Y) returns the remainder after X is divided by Y. The mapping modulo 1 realizes the normalization of the system parameter range.

[0033] S1_1_3. Discard the first 800 iterations of the chaotic sequence obtained, and calculate the chaotic pseudo-random sequence X=(x1, x2, x3…, x4) with a length and width of m×n through parallel computing. m×n ) and Y=(y1,y2,y3…,y m×n );

[0034] S1_1_4, X m×n With Ym×n Interleave sorting, intercept the first m×n long sequence, and obtain the final chaotic pseudo-random sequence Q=(x1,y1,x2,y2,…);

[0035] S1_1_5, respectively take Q m×n The mth, nth, (m+n)th, and (m×n)th elements in the sequence form the 2D-Logistic key D = (Q m ,Q n ,Q m+n ,Q m×n ), as the initial key for subsequent key iterations;

[0036] S1_1_6. The above steps are completed.

[0037] Func1.2, RANSAC-circle key

[0038] The algorithm used by the key is RANSAC, which estimates the parameters of the selected mathematical model from a set of external observation data sets through an iterative method, and obtains the final result by repeatedly iterating and selecting a set of random subsets in the data set. The algorithm has a certain degree of uncertainty, and a reasonable result will be obtained with a certain probability. It is consistent with the characteristics of avalanche and randomness in chaotic systems, and the calculation speed is fast. Selecting a model with low error rate and high accuracy will sufficiently reduce the calculation error rate, which is suitable for digital image encryption. The specific steps are described as follows:

[0039] S1_2_1. Obtain 2D-Logistic key D = (Q m ,Q n ,Q m+n ,Q m×n );

[0040] S1_2_2, import the chaotic pseudo-random sequence Q = (x1, y1, x2, y2, ...) generated in Func 1.1 as the external observation data set;

[0041] S1_2_3, initialize the selected model to be a circle;

[0042] S1_2_4, import the 2D-Logistic key D into the parameter set required by the RANSAC algorithm, which is shown in formula E3;

[0043]

[0044] Among them, formula E3 is the RANSAC parameter generation formula; r = (r1, r2, r3) is the radius set of the circle model, x a is the noise coefficient, x n=(x1,x2,x3) is the number of data points in the circle model, x c =(x c1 , x c2 , x c3 ) and y c =(y c1 ,y c2 ,y c3 ) are the horizontal and vertical coordinate sets of the center coordinates of the circle model respectively; Q m , Q n , Q m+n , Q m×n All are elements in key D;

[0045] S1_2_5, bringing the parameter set into the RANSAC-circle algorithm shown in Formula E4 for iteration to form a pseudo-random sequence Rc=(Rc1, Rc2…);

[0046]

[0047] Wherein, Formula E4 is the RANSAC equation; rand(m,n) is a function for generating pseudo-random numbers of uniform distribution, and the generation interval is (m,n); randn(m,n) is a function for generating pseudo-random numbers of standard normal distribution, and the generation interval is (m,n) with mean 0 and variance 1; pi is the pi constant.

[0048] S1_2_6, take the mth, nth, (m+n)th, and (m×n)th elements in the Rc sequence respectively to form the RANSAC-circle key R = (R m ,R n ,R m+n ,R m×n ), as the initial key for subsequent key iterations;

[0049] S1_2_7, the steps are completed.

[0050] Func1.3, superLorenz key

[0051] This key is based on the superLorenz mapping and is used as the final key of the "key string", that is, the key used to generate the sequence of the overall chaotic cipher generator. The specific steps are described as follows:

[0052] S1_3_1. Obtain the RANSAC-circle key R = (R m ,R n ,R m+n ,R m×n );

[0053] S1_3_2, substitute the key R into the superLorenz mapping of E5 and iterate m×n+800 times, adding the periodic perturbation of E6 every 3000 iterations;

[0054]

[0055] E6:x0=R m +h×sin(R n )

[0056] Among them, Formula E5 is the superLorenz mapping equation, Formula E6 is the periodic sinusoidal perturbation formula; R m , R n is the element in key R, is the state variable of the mapping, a, b, c, r are system parameters, when a=10, b=8 / 3, c=28, r∈[-1.52,-0.06], the system is in a chaotic state;

[0057] S1_3_3. Discard the first 800 iterations of the obtained chaotic sequence, and obtain four chaotic pseudo-random sequences X=(x1,x2,…), Y=(y1,y2,…), Z=(z1,z2,…), and M=(m1,m2,…) of length m×n through parallel calculation.

[0058] S1_3_4, put the sequences X, Y, Z, M into the parallel pool;

[0059] S1_3_5. Convert the decimals in each sequence in the parallel pool into 16-bit binary numbers;

[0060] S1_3_6, sort the even-numbered digits of the sequence value in step S1_3_5 in positive order and put them in the first place;

[0061] S1_3_7. Put the odd digits of the sequence value in step S1_3_6 at the end in reverse order, reorganize into a binary number, and then convert into a decimal number.

[0062] S1_3_8. After the above bit rearrangement, a new chaotic pseudo-random sequence Xs=(xs1, xs2, ...), Ys=(ys1, ys2, ...), Zs=(zs1, zs2, ...), Ms=(ms1, ms2, ...) is generated;

[0063] S1_3_9, the steps are completed.

[0064] Func1.4, Generate encrypted random matrix

[0065] The four chaotic pseudo-random sequences generated by the above steps are converted into four random matrices of the same size as the plaintext image for subsequent encryption operations. The specific steps are described as follows:

[0066] S1_4_1, randomly select four 8-bit integers, c1, c2, c3, c4, whose interval is [0, 255];

[0067] S1_4_2. Convert the sequences Xs, Ys, Zs, and Ms in parallel according to formula E7 to obtain encrypted random matrices X(m,n), Y(m,n), Z(m,n), and M(m,n) of corresponding sizes of m×n.

[0068]

[0069] Wherein, formula E7 is the chaotic sequence generation equation; floor(X) is an integer not greater than X, i = 1, 2, ..., m, j = 1, 2, ..., n;

[0070] S1_4_3, the steps are completed.

[0071] 2. Digital image encryption of network approval documents

[0072] The encryption process diagram is as follows Figure 1 As shown. First, receive the network electronic document image P and divide it into block images in units of 2×2, and put it into a parallel pool for parallel encryption operations. If the length and width are not divisible, they are corrected with 0 pixel values. The corrected size is m×n. Taking the first block image P1 as an example, P1 is preprocessed to obtain the preprocessed image P', and the remaining image blocks are processed in parallel. Then, each block of the image is parallelly subjected to plaintext-independent forward diffusion, plaintext-associated scrambling, and plaintext-independent backward diffusion, and finally the electronic document ciphertext image C is obtained. The specific steps are described as follows:

[0073] Func2.1, digital image preprocessing

[0074] The specific steps are described as follows:

[0075] The plaintext block image P1 and the random matrix X(m / 2,n / 2) are subjected to bitwise XOR processing using formula E8 to obtain the preprocessed image P';

[0076] E8:P'(i,j)=P(i,j)⊕X(i,j)

[0077] Where A⊕B represents an XOR operation on A and B, i=1,2,…,m / 2, j=1,2,…,n / 2;

[0078] Func 2.2, plaintext-independent forward diffusion

[0079] The preprocessed image P' and the random matrix Y(m / 2,n / 2) are forward diffused into the matrix A. The specific steps are described as follows:

[0080] S2_2_1, convert P'(1,1) into A(1,1) by formula E9;

[0081] E9:A(1,1)=(P(1,1)+Y(1,1)+c1+c2)mod 256

[0082] Among them, c1 and c2 are random parameters in S1_4_1;

[0083] S2_2_2, convert P'(i,1) into A(i,1) by formula E10;

[0084] E10:A(i,1)=(P(i,1)+Y(i,1)+A(i-1,1))mod 256

[0085] Where i = 1, 2, ..., m / 2;

[0086] S2_2_3, convert P'(1,j) into A(1,j) by formula E11;

[0087] E11:A(1,j)=(P(1,j)+Y(1,j)+A(1,j-1))mod 256

[0088] Where, j = 1, 2, ..., n / 2;

[0089] S2_2_4, convert P'(i,j) into A(i,j) by formula E12;

[0090] E12:A(i,j)=(P(i,j)+Y(i,j)+A(i,j-1)+A(i-1,j))mod 256

[0091] Where, i = 1, 2, ..., m / 2, j = 1, 2, ..., n / 2;

[0092] S2_2_5. The above steps are completed.

[0093] Func2.3, plaintext association scrambling

[0094] Swap the positions of the corresponding pixels A(i,j) and A(m / 2,n / 2) in matrix A, where i=1,2,…,m / 2 and j=1,2,…,n / 2.

[0095] The specific steps are described as follows:

[0096] S2_3_1. Calculate the sum of the pixel values ​​of all the points in the row where the pixel point A(i,j) is located using formula E13, and record it as ri;

[0097] E13: ri = sum(A(i, 1 to n / 2)) – A(i, j)

[0098] Where sum(A(i, 1 to n)) represents returning the sum of elements in the i-th row of matrix A;

[0099] S2_3_2: Calculate the sum of pixel values of all points in the column where the pixel point A(i, j) is located according to formula E14, and denote it as ci;

[0100] E14: ci = sum(A(1 to m / 2, j)) – A(i, j)

[0101] S2_3_3: Calculate the value of the coordinate (mi, ni) according to formula E15;

[0102]

[0103] S2_3_4: If m = i or n = j, the positions of A(m / 2, n / 2) and A(i, j) remain unchanged; otherwise, they are swapped, and A(i, j) is circularly left-shifted according to formula E16 based on the pixel values of the lower three bits of A(m, n);

[0104] E16: A(i, j) = A(i, j) <<< (A(m / 2, n / 2) & 0x7)

[0105] Where A <<< B returns the result of circularly left-shifting A by B bits; & is the binary AND operation, and 0x7 is the decimal representation of binary 111.

[0106] S2_3_5: After the above steps, finally obtain the scrambled image B of image A;

[0107] S2_3_6: The above steps are completed.

[0108] Func2.4: Plaintext-independent backward diffusion

[0109] Perform plaintext-independent backward diffusion on the image B obtained in Func2.3 and the matrix M in Func1.4 to obtain image C. The specific steps are described as follows:

[0110] S2_4_1: Convert B(m / 2, n / 2) to C(m / 2, n / 2) according to formula E17;

[0111] E17: C(m / 2, n / 2) = (B(m / 2, n / 2) + M(m / 2, n / 2) + c3 + c4) mod 256

[0112] Where c3 and c4 are the random parameters in S1_4_1;

[0113] S2_4_2, transform B(m / 2,j) into C(m / 2,j) by formula E18;

[0114] E18:C(m / 2,j)=(B(m / 2,j)+M(m / 2,j)+C(m / 2,j+1))mod 256

[0115] Where, j = n / 2-1, n / 2-2…3, 2;

[0116] S2_4_3, transform B(i,n / 2) into C(i,n / 2) using formula E19;

[0117] E19:C(i,n / 2)=(B(i,n / 2)+M(n / 2)+C(i+1,n / 2))mod 256

[0118] Where i = m / 2-1, m / 2-2…3, 2;

[0119] S2_4_4, transform B(i,j) into C(i,j) using formula E20;

[0120] E20:C(i,j)=(B(i,j)+C(i+1,j)+C(i,j+1)+M(i,j))mod 256

[0121] Where, i = m / 2-1, m / 2-2…2,1, j = n / 2-1, n / 2-2…,2,1;

[0122] S2_4_5. After the above steps, a network electronic document ciphertext image C is obtained;

[0123] S2_4_6. The encryption process ends.

[0124] (III) Beneficial effects

[0125] The beneficial effects of the present invention are:

[0126] 1. The chaotic password generator is constructed in the form of a "key string" of three sets of iterative keys. The generator generates a chaotic sequence in the form of hyperchaotic mapping and RANSAC stacking in parallel. The sequence is used as the input sequence of the subsequent encryption process to encrypt the network electronic document image. In the process, 2D-Logistic key, RANSAC-circle key, and superLorenz key are iteratively generated one by one, and the anti-chaotic degradation characteristics are formed through periodic sinusoidal perturbations. The chaotic pseudo-random sequence is calculated by bit rearrangement and parallel calculation. This method can enhance the sensitivity of the image, improve the key sensitivity, reduce the calculation time required to generate the sequence for encrypting large pictures, expand the key space, and effectively resist classic attacks.

[0127] 2. The encryption process uses a pre-processing-diffusion-scrambling-diffusion architecture to perform block-by-block parallel encryption on network electronic document images. This method can effectively improve the calculation speed of the encryption process, increase the sensitivity of plaintext, and increase information entropy. It can effectively resist attacks such as statistical attacks and differential attacks, and ensure the security of cross-transmission of document data in the internal network.

[0128] 3. Since all formulas in the encryption process can be reversed, the decryption process is the reverse process of the encryption process. The present invention solves the problem of data inconsistency caused by calculation of different word lengths in network transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0129] Figure 1 This is a module flow chart of the present invention.

[0130] Figure 2 This is a schematic diagram of the parallel encryption framework described in the present invention.

[0131] Figure 3 These are sample images of encryption / decryption with different lengths and widths.

[0132] Figure 4 It is the histogram of plaintext and ciphertext grayscale images.

[0133] Figure 5(a) is a graph showing the correlation between adjacent pixels of a plaintext image.

[0134] Figure 5(b) is a graph showing the correlation between adjacent pixels of a ciphertext image.

[0135] Figure 6 This is a qualitative analysis diagram of key sensitivity. DETAILED DESCRIPTION

[0136] In order to explain the encryption technology scheme of the present invention more clearly and completely, the following is combined with Figure 1 The module flow chart, Figure 2 The parallel encryption framework schematic diagram and examples are provided to further illustrate the present invention, but are not intended to limit the scope of the invention.

[0137] Example 1: Constructing a Chaotic Code Generator

[0138] The overall encryption process requires a pseudo-random chaotic sequence generated by a chaotic cipher generator for scrambling and diffusion operations in the encryption process. The chaotic cipher generator is constructed from a "key string", which consists of three sets of mutually iterative "keys": 2D-Logistic key, RANSAC-circle key, and superLorenz key. The four sequences generated by the chaotic cipher generator are used for preprocessing, scrambling, diffusion and other operations in the encryption process;

[0139] P represents the sample electronic document image. If the length and width are not divisible, they are corrected with 0 pixel value. The corrected image size is 800×800. The key generation steps are as follows:

[0140] Func1.1, 2D-Logistic key

[0141] The key is based on the two-dimensional Logistic (i.e. 2D-Logistic) mapping and serves as the initial iteration key of the "key string". The specific steps are described as follows:

[0142] S1_1_1, input the initial parameters of the two-dimensional Logistic chaotic map u1=3.99, u2=3.97;

[0143] S1_1_2, Substitute the initial parameters u1 and u2 into the two-dimensional Logistic chaotic mapping equation E1 and iterate 640800 times, adding the periodic disturbance of equation E2 every 3000 iterations;

[0144] S1_1_3. Discard the first 800 iterations of the chaotic sequence, and obtain the chaotic pseudo-random sequences X=(0.9402, 0.2166, 0.9452, 0.6115, …) and Y=(0.3026, 0.5229, 0.9570, 0.1561, …) with a length of 640000 through parallel calculation.

[0145] S1_1_4, X m×n With Y m×n Interlaced sorting, intercepting the first 640000 long sequences, and obtaining the final chaotic pseudo-random sequence Q = (0.9402, 0.3026, 0.2166, 0.5229, 0.9452, 0.9570, 0.6115, 0.1561, ...);

[0146] S1_1_5, respectively take Q m×n The 400th, 800th, 1600th, and 640000th elements in the sequence form the 2D-Logistic key D = (0.7246, 0.1262, 0.8565, 0.6267), which is used as the initial key for subsequent key iterations;

[0147] S1_1_6. The above steps are completed.

[0148] Func1.2, RANSAC-circle key

[0149] The specific steps to generate the RANSAC-circle key are described as follows:

[0150] S1_2_1. Get 2D-Logistic key D = (0.7246, 0.1262, 0.8565, 0.6267);

[0151] S1_2_2, import the chaotic pseudo-random sequence Q generated in Func1.1 as the external observation data set;

[0152] S1_2_3, select the model as a circle;

[0153] S1_2_4, substitute the 2D-Logistic key D into the RANSAC parameter generation formula E3 to generate a parameter set;

[0154] S1_2_5, bring the parameter set into the RANSAC-circle algorithm shown in RANSAC equation E4 for iteration, and finally form a pseudo-random sequence Rc=(0.8701, 0.6548, 0.5138, 0.0974, ...);

[0155] S1_2_6, take the 400th, 800th, 1600th, and 640000th elements in the Rc sequence respectively to form the RANSAC-circle key R = (0.9265, 0.5527, 0.7160, 0.2270, ...), which is used as the initial key for subsequent key iterations;

[0156] S1_2_7, the steps are completed.

[0157] Func1.3, superLorenz key

[0158] This key is based on the superLorenz map and serves as the final key of the "key string", that is, the key used to generate the sequence of the overall chaotic cipher generator.

[0159] The specific steps are described as follows:

[0160] S1_3_1, obtain the RANSAC-circle key R;

[0161] S1_3_2, substitute the key R into the superLorenz mapping equation E5 and iterate 640800 times, and achieve the purpose of anti-chaotic degradation through periodic sinusoidal perturbation formula E6 every 3000 iterations;

[0162] S1_3_3. Discard the first 800 iterations of the obtained chaotic sequence, and obtain four chaotic pseudo-random sequences of length 640,000 through parallel calculation:

[0163] X=(0.0855,0.6457,0.5832,0.7404,…);

[0164] Y=(0.3038,0.6123,0.3662,0.5848,…);

[0165] Z=(0.0534,0.3045,0.4606,0.1452,…);

[0166] M=(0.6456,0.4910,0.2417,0.7389,…);

[0167] S1_3_4, put the sequences X, Y, Z, M into the parallel pool;

[0168] S1_3_5. Rearrange the bits of the elements in the sequence respectively: convert the decimal into a 16-bit binary number, sort its even bits in ascending order and put them in the first place, and put its odd bits in descending order at the end, reconstruct the binary number, and then convert it into a decimal number.

[0169] S1_3_6. After the above bit rearrangement, a new chaotic pseudo-random sequence Xs=(0.6899, 0.2631, 0.2553, ...), Ys=(0.0815, 0.9605, 0.9568, ...), Zs=(0.9553, 0.7060, 0.4477, ...), Ms=(0.5131, 0.7117, 0.2114, ...) is generated;

[0170] S1_3_7, the steps are completed.

[0171] Func1.4, Generate encrypted random matrix

[0172] The four chaotic pseudo-random sequences generated by the above steps are converted into four random matrices with the same size as the plaintext image for subsequent encryption operations.

[0173] The specific steps are described as follows:

[0174] S1_4_1, randomly select four eight-bit integers, which are c1=12, c2=64, c3=48, c4=142;

[0175] S1_4_2. Parallel transform the sequences Xs, Ys, Zs, and Ms according to the chaotic sequence generation equation E7 to obtain random matrices X(800,800), Y(800,800), Z(800,800), and M(800,800) of corresponding size of 640,000.

[0176] S1_4_3, the above steps are completed.

[0177] Example 2. Digital image encryption

[0178] First, the network electronic document image P is divided into blocks of images in units of 2×2, and placed in a parallel pool for parallel encryption operations. If the length and width are not divisible, they are corrected with 0 pixel values, and the corrected size is 640000. Taking the first block image P1 as an example, P1 is preprocessed to obtain the preprocessed image P'. Then, each block image is parallelly subjected to plaintext-independent forward diffusion, plaintext-associated scrambling, and plaintext-independent backward diffusion, and finally the electronic document ciphertext image C is obtained.

[0179] The specific steps are described as follows:

[0180] Func2.1, digital image preprocessing

[0181] The specific steps are described as follows:

[0182] Perform bit-level XOR processing on the plaintext block image P1 and the random matrix X(400,400) to obtain the preprocessed image P';

[0183] Func2.2, plaintext-independent forward diffusion

[0184] The preprocessed image P' and the random matrix Y(400,400) are forward diffused and transformed into the matrix A. The specific steps are described as follows:

[0185] S2_2_1, convert P'(1,1) into A(1,1), P'(i,1) into A(i,1), P'(1,j) into A(1,j), P'(i,j) into A(i,j);

[0186] S2_2_2, the above steps are completed.

[0187] Func2.3, plaintext association scrambling

[0188] Swap the positions of the corresponding pixel points A(i,j) of matrix A with A(400,400), where i=1,2,…,400 and j=1,2,…,400.

[0189] The specific steps are described as follows:

[0190] S2_3_1. Calculate the sum of the pixel values ​​of all the points in the row where the pixel point A(i,j) is located, denoted as ri;

[0191] S2_3_2, calculate the sum of the pixel values ​​of all the points in the column where the pixel point A(i,j) is located, denoted as ci;

[0192] S2_3_3, calculate the value of coordinates (mi, ni) through sequences Z and M;

[0193] S2_3_4. If m=i or n=j, the positions of A(m / 2,n / 2) and A(i,j) remain unchanged, otherwise they are replaced, and A(i,j) is circularly shifted left according to the pixel values ​​of the lower three binary digits of A(m,n);

[0194] S2_3_5, finally obtaining the image B after scrambling the image A;

[0195] S2_3_6. The above steps are completed.

[0196] Func2.4, plaintext-independent back diffusion

[0197] Image C is obtained by performing plaintext-independent back diffusion on the image B obtained in Func2.3 and the matrix M in 1.4.

[0198] The specific steps are described as follows:

[0199] S2_4_1. Convert B(m / 2,n / 2) to C(m / 2,n / 2), B(m / 2,j) to C(m / 2,j), B(i,n / 2) to C(i,n / 2), and B(i,j) to C(i,j);

[0200] S2_4_2, finally get the ciphertext image C, such as Figure 3 This is a sample diagram for encryption and decryption;

[0201] S2_4_3. The encryption process ends.

[0202] The performance of the image encryption algorithm of the above example of the present invention is analyzed below.

[0203] 1. Histogram analysis of plaintext and ciphertext grayscale images

[0204] like Figure 4 The following are examples of histograms of plaintext and ciphertext grayscale images of the present invention, and the corresponding encrypted images are sample document 1, sample document 2, and sample document 3. As shown in the figure, the encrypted grayscale image histogram is smooth and flat, which can effectively resist the statistical attack method of plaintext pixel values ​​by illegal attackers and achieve good encryption effect.

[0205] 2. Correlation Analysis

[0206] 300 adjacent pixels are randomly obtained in the plaintext and ciphertext grayscale images to analyze the pixel correlation of the corresponding images, and the adjacent pixel correlation diagram is drawn according to the correlation coefficient calculation formula, as shown in Figure 5(a) and Figure 5(b).

[0207]

[0208]

[0209]

[0210]

[0211] Among them, N is N pairs of adjacent pixels, and its gray value is (u i ,v i ), i = 1, 2, ..., N, cov(u, v) is the covariance, E(u) is the mathematical expectation. After calculation, the correlation coefficients of neighboring pixels in the horizontal direction, vertical direction, diagonal direction, and anti-diagonal direction are obtained, as shown in Table 1. If the corresponding direction correlation coefficient approaches 1, the correlation is strong, and if it is less than or equal to 0, the correlation is weak.

[0212] As shown in Table 1, it shows that the correlation coefficient of the encryption algorithm of the present invention is small enough to effectively reduce the pixel correlation and achieve a good encryption effect.

[0213] Table 1. Image neighboring pixel correlation coefficient table

[0214]

[0215] 3. Plaintext sensitivity analysis

[0216] According to the following formula, two parameters, NPCR and UACI, are used to analyze the plaintext sensitivity of the present invention.

[0217]

[0218]

[0219] By using the encryption method of the present invention, two sample document images with similar visual effects are constructed, and the original image parameters NPCR=100%, UACI=32.1895% are obtained, and the corresponding image parameters NPCR=100%, UACI=38.8911% with similar effects are obtained.

[0220] It can be concluded that the encryption method proposed by the present invention has strong plaintext sensitivity and can effectively resist differential attacks by illegal attackers.

[0221] 4. Key sensitivity analysis

[0222] A qualitative analysis is conducted on the key sensitivity of the encryption method proposed in the present invention, and the difference image of the grayscale image of the corresponding sample document after a slight change in the key is observed.

[0223] The analysis results are as follows Figure 6 As shown in the figure, it can be seen that after a slight change in the key, the difference image corresponding to the two ciphertext images is still a noise-like image intuitively, so the encryption method of the present invention has strong key sensitivity.

[0224] Finally, it should be noted that the above examples are only used to illustrate the technical method of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above examples, a person skilled in the art should understand that the technical solutions described in the above examples can still be modified, or some of the technical features can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the examples of the present invention.

Claims

1. A method for parallel encryption of images in network office approval documents based on hyperchaos, characterized by: (1) The encryption process uses a chaotic cipher generator to generate a pseudo-random chaotic sequence to achieve scrambling and diffusion operations; the chaotic cipher generator is composed of a key string, which is composed of three sets of mutually iterative keys: 2D-Logistic key, RANSAC-circle key, and superLorenz key; the four sequences generated by the chaotic cipher generator are used for preprocessing, scrambling, and diffusion operations in the encryption process; (2) Receive the network electronic document image P; divide it into blocks of 2×2 units, store it in a parallel pool for parallel encryption processing, and if the length and width m, n are not divisible, they are corrected with 0 pixel values ​​to become m×n after correction; the first block of image is P1, and P1 is preprocessed to obtain the preprocessed image P', and the remaining image blocks are processed in parallel. For each block of image, plaintext-independent forward diffusion, plaintext-associated scrambling, and plaintext-independent backward diffusion are performed in parallel to finally obtain the electronic document ciphertext image C; A hyperchaotic image parallel encryption method in network office approval documents, specifically comprising: Part_1. Constructing a Chaotic Code Generator The key generation steps are as follows: Func1.1, 2D-Logistic key The key is based on two-dimensional Logistic, i.e. 2D-Logistic mapping, as the initial iteration key of the key string. The specific steps are described as follows: S1_1_1, input the initial parameters u1 and u2 of the two-dimensional Logistic chaotic map; S1_1_2, substitute into E1 and iterate m×n+800 times, adding the periodic perturbation of E2 every 3000 iterations; E2:x0=x0+h×sin(y0) Among them, E1 is the two-dimensional Logistic chaos mapping equation, E2 is the periodic sinusoidal perturbation formula; x0 and y0 are the initial state variables of the mapping, x i ,y i is the state variable of the mapping, u1 and u2 are system parameters, h is the mapping step size, when the system parameter value range is (0,4], the system is in a chaotic state; X mod Y returns the remainder after X is divided by Y; the mapping modulo 1 realizes the normalization of the system parameter range; S1_1_3. Discard the first 800 iteration results in the obtained chaotic sequence, and calculate the chaotic pseudo-random sequence X=(x1, x2, x3…, x m×n ) and Y=(y1,y2,y3…,y m×n ); S1_1_4, X m×n With Y m×n Interleave sorting, intercept the first m×n long sequence, and obtain the final chaotic pseudo-random sequence Q=(x1,y1,x2,y2,…); S1_1_5, respectively take Q m×n The mth, nth, (m+n)th, and (m×n)th elements in the sequence form the 2D-Logistic key D = (Q m ,Q n ,Q m+n ,Q m×n ), as the initial key for subsequent key iterations; S1_1_6, the steps are completed; Func1.2, RANSAC-circle key The key is based on the RANSAC algorithm, and the specific steps are described as follows: S1_2_1. Obtain 2D-Logistic key D = (Q m ,Q n ,Q m+n ,Q m×n ); S1_2_2, import the chaotic pseudo-random sequence Q = (x1, y1, x2, y2, ...) generated in Func 1.1 as the external observation data set; S1_2_3, initialize the selected model to be a circle; S1_2_4, import the 2D-Logistic key D into the parameter set required by the RANSAC algorithm, which is formula E3; Among them, formula E3 is the RANSAC parameter generation formula; r = (r1, r2, r3) is the radius set of the circle model, x a is the noise coefficient, x n =(x1,x2,x3) is the number of data points in the circle model, x c =(x c1 , x c2 , x c3 ) and y c =(y c1 ,y c2 ,y c3 ) are the horizontal and vertical coordinate sets of the center coordinates of the circle model respectively; Q m , Q n , Q m+n , Q m×n All are elements in key D; S1_2_5, bring the parameter set into the RANSAC-circle algorithm of formula E4 for iteration to form a pseudo-random sequence Rc=(Rc1, Rc2…); Wherein, formula E4 is the RANSAC equation; rand(m,n) is a function for generating uniformly distributed pseudo-random numbers, and the generation interval is (m,n); randn(m,n) is a function for generating standard normally distributed pseudo-random numbers, and the generation interval is (m,n) with a mean of 0 and a variance of 1; pi is the pi constant; S1_2_6, take the mth, nth, (m+n)th, and (m×n)th elements in the Rc sequence respectively to form the RANSAC-circle key R = (R m ,R n ,R m+n ,R m×n ), as the initial key for subsequent key iterations; S1_2_7, the steps are completed; Func1.3, superLorenz key The specific steps are described as follows: S1_3_1. Get the RANSAC-circle key R = (R m ,R n ,R m+n ,R m×n ); S1_3_2, substitute the key R into the superLorenz mapping of E5 and iterate m×n+800 times, adding the periodic perturbation of E6 every 3000 iterations; E6:x0=R m +h×sin(R n ) Among them, E5 is the superLorenz mapping equation, E6 is the periodic sinusoidal perturbation formula; R m , R n is the element in key R, is the state variable of the mapping, a, b, c, r are system parameters, initialized with a=10, b=8 / 3, c=28, r∈[-1.52,-0.06]; S1_3_3. Discard the first 800 iteration results in the obtained chaotic sequence, and obtain four chaotic pseudo-random sequences X=(x1,x2,…), Y=(y1,y2,…), Z=(z1,z2,…), M=(m1,m2,…) of length m×n through parallel calculation; S1_3_4, put the sequences X, Y, Z, M into the parallel pool; S1_3_5. Convert the decimals in each sequence in the parallel pool into 16-bit binary numbers; S1_3_6, sort the even-numbered digits of the sequence value in step S1_3_5 in positive order and put them in the first place; S1_3_7, put the odd digits of the sequence value in step S1_3_6 at the end in reverse order, reorganize into a binary number, and then convert into a decimal number; S1_3_8. After the bit rearrangement, a new chaotic pseudo-random sequence Xs=(xs1, xs2, ...), Ys=(ys1, ys2, ...), Zs=(zs1, zs2, ...), Ms=(ms1, ms2, ...) is generated; S1_3_9, the steps are completed; Func1.4, Generate encrypted random matrix The four chaotic pseudo-random sequences are converted into four random matrices of the same size as the plaintext image to achieve encryption operation; the specific steps are described as follows: S1_4_1, randomly select four 8-bit integers, c1, c2, c3, c4, whose interval is [0, 255]; S1_4_2. Convert the sequences Xs, Ys, Zs, and Ms in parallel according to formula E7 to obtain encrypted random matrices X(m,n), Y(m,n), Z(m,n), and M(m,n) of corresponding size m×n; Wherein, formula E7 is the chaotic sequence generation equation; floor(X) is an integer not greater than X, i = 1, 2, ..., m, j = 1, 2, ..., n; S1_4_3, the steps are completed; Part_1 description ends; Part_2: Digital Image Encryption of Network Approvals Func2.1, digital image preprocessing The specific steps are described as follows: The plaintext block image P1 and the random matrix X(m / 2,n / 2) are subjected to bitwise XOR processing using formula E8 to obtain the preprocessed image P'; E8:P'(i,j)=P(i,j)⊕X(i,j) Where A⊕B represents an XOR operation on A and B, i=1,2,…,m / 2, j=1,2,…,n / 2; Func 2.2, plaintext-independent forward diffusion The preprocessed image P' and the random matrix Y(m / 2,n / 2) are forward diffused into the matrix A; The specific steps are described as follows: S2_2_1, convert P'(1,1) into A(1,1) by formula E9; E9:A(1,1)=(P(1,1)+Y(1,1)+c1+c2)mod 256 Among them, c1 and c2 are random parameters in S1_4_1; S2_2_2, convert P'(i,1) into A(i,1) by formula E10; E10: A(i, 1) = (P(i, 1) + Y(i, 1) + A(i - 1, 1)) mod 256 where i = 1, 2, …, m / 2; S2_2_3. Convert P’(1, j) to A(1, j) through formula E11; E11: A(1, j) = (P(1, j) + Y(1, j) + A(1, j - 1)) mod 256 where j = 1, 2, …, n / 2; S2_2_4. Convert P’(i, j) to A(i, j) through formula E12; E12: A(i, j) = (P(i, j) + Y(i, j) + A(i, j - 1) + A(i - 1, j)) mod 256 where i = 1, 2, …, m / 2, j = 1, 2, …, n / 2; S2_2_5. The above steps are completed; Func2.

3. Plaintext-related scrambling Swap the positions of the corresponding pixel points A(i, j) and A(m / 2, n / 2) in matrix A, where i = 1, 2, …, m / 2, j = 1, 2, …, n / 2; The specific steps are described as follows: S2_3_1. Calculate the sum of pixel values of all points in the row where pixel point A(i, j) is located according to formula E13, denoted as ri; E13: ri = sum(A(i, 1 to n / 2)) – A(i, j) where sum(A(i, 1 to n)) represents the sum of elements in the i-th row of matrix A; S2_3_2. Calculate the sum of pixel values of all points in the column where pixel point A(i, j) is located according to formula E14, denoted as ci; E14: ci = sum(A(1 to m / 2, j)) – A(i, j) S2_3_3. Calculate the value of coordinates (mi, ni) according to formula E15; S2_3_4. If m = i or n = j, the positions of A(m / 2, n / 2) and A(i, j) remain unchanged, otherwise they are swapped, and A(i, j) is circularly left-shifted according to formula E16 based on the pixel values of the lower three bits of A(m, n); E16: A(i, j) = A(i, j) <<< (A(m / 2, n / 2) & 0x7) where A <<< B returns the result of circularly left-shifting A by B bits; & is the binary AND operation, and 0x7 is the decimal representation of binary 111; S2_3_5. Obtain the scrambled image B of image A; S2_3_6. The above steps are completed; Func2.

4. Plaintext-unrelated backward diffusion The image B obtained in Func2.3 and the matrix M in Func1.4 are subjected to plaintext-unrelated backward diffusion to obtain image C; The specific steps are described as follows: S2_4_1. Convert B(m / 2, n / 2) to C(m / 2, n / 2) according to formula E17; E17: C(m / 2, n / 2) = (B(m / 2, n / 2) + M(m / 2, n / 2) + c3 + c4) mod 256 where c3 and c4 are random parameters in S1_4_1; S2_4_2. Convert B(m / 2, j) to C(m / 2, j) according to formula E18; E18:C(m / 2,j)=(B(m / 2,j)+M(m / 2,j)+C(m / 2,j+1))mod 256 Where, j = n / 2-1, n / 2-2…3, 2; S2_4_3, transform B(i,n / 2) into C(i,n / 2) using formula E19; E19:C(i,n / 2)=(B(i,n / 2)+M(n / 2)+C(i+1,n / 2))mod 256 Where i = m / 2-1, m / 2-2…3, 2; S2_4_4, transform B(i,j) into C(i,j) using formula E20; E20:C(i,j)=(B(i,j)+C(i+1,j)+C(i,j+1)+M(i,j))mod 256 Where, i = m / 2-1, m / 2-2…2,1, j = n / 2-1, n / 2-2…,2,1; S2_4_5, obtaining the encrypted image C of the network electronic official document; S2_4_6, the encryption process ends; End of Part_2 description.