A Medical Image Encryption Method Based on Cellular Automata and Embedded ROI Positions

Abstract

The present invention discloses a medical image encryption method based on cellular automata and the embedding of ROI positions. First, the region of interest (ROI) is selected according to the block mean threshold and the corresponding marking information is recorded. The obtained ROI is reorganized into a rectangle with the closest length and width, and then binary conversion is performed to convert it into multiple bit planes with the same configuration. These bit planes are grouped in pairs as the initial input of the cellular automata. The local rules are calculated based on the key and the hash value of the plaintext image, and encryption is performed using the number of iteration rounds of the cellular automata. A preliminary ciphertext image is formed by combining the region of non-interest (RONI). The marking information of the ROI is embedded into the ciphertext image using the histogram shifting technique, and finally the ciphertext image is output. The image encryption efficiency of this method is high.

Description

Technical Field

[0001] The present invention relates to a medical image encryption method based on cellular automata and the embedded ROI position, belonging to the technical field of information encryption. Background Art

[0002] Medical images are important auxiliary means for medical diagnosis and clinical treatment. However, without any security protection, directly transmitting medical images over the Internet may cause malicious attacks and privacy leaks. Different from natural images, medical images contain a large amount of black background. For privacy protection, the background of medical images does not need to be encrypted. Therefore, the encryption scheme based on the region of interest is particularly suitable for medical image encryption, which can improve the encryption efficiency without sacrificing security.

[0003] However, most of the existing ROI-based encryption schemes take up a large amount of space and time to transmit the marker information of the ROI, and have poor parallelism. Summary of the Invention

[0004] To solve the above problems, the present invention provides a medical image encryption method based on cellular automata and the embedded ROI position.

[0005] The present invention adopts the following technical solutions to solve the above technical problems:

[0006] A medical image encryption method based on cellular automata and the embedded ROI position, the method comprising:

[0007] Step 1, divide the plaintext image into blocks, extract the region of interest ROI and the region of non-interest RONI in the original image based on the mean value of each block, and record the corresponding ROI marker information;

[0008] Step 2, calculate eight groups of balanced rules of the cellular automata according to the key and the hash value of the plaintext image;

[0009] Step 3, reorganize the region of interest into a rectangle with the smallest difference between the length and width, and use the cellular automata to perform iterative encryption;

[0010] Step 4, embed the binary-form diagnostic information into the region of non-interest RONI through the histogram shifting technique;

[0011] Step 5, combine the encrypted region of interest ROI and the region of non-interest RONI with the embedded information to form a preliminary ciphertext image C1;

[0012] Step 6, embed the ROI marker information in Step 1 into the preliminary ciphertext image C1 through the histogram shifting technique to obtain the final encrypted image.

[0013] Further, step 1 specifically includes:

[0014] Divide the plaintext image into blocks, calculate the pixel mean of each block respectively, and set up a matrix H. If the pixel mean of the block located at the i-th row and j-th column in the plaintext image is greater than the preset block mean threshold, then the element H(i,j) at the i-th row and j-th column in H is 1; otherwise, H(i,j) = 0.

[0015] Scan the matrix H row by row. If there is no element with a value of 1 in a certain row, no processing is done and the next row is scanned; if there are elements with a value of 1, then all the elements between the first element with a value of 1 and the last element with a value of 1 in this row are set to 1, and the row number of this row and the column numbers corresponding to the first element with a value of 1 and the last element with a value of 1 are converted into binary to generate a triple (if there is only one element with a value of 1 in this row, after converting the column number corresponding to the element with a value of 1 into binary, record both sides to generate a triple); after the scanning is completed, a marked matrix H’ and marked information composed of all triples are obtained.

[0016] If H’(i,j) × 1, then the corresponding block belongs to the region of interest ROI; otherwise, it belongs to the region of non-interest RONI.

[0017] Further, the calculation method of each group of balance rules in step 2 specifically includes:

[0018] Divide the 512-bit binary key into eight groups, with each group containing 64-bit binary numbers; for every eight bits in the 64-bit binary numbers of each group, convert them into a decimal number, obtaining 8 decimal numbers in the range (0, 255). Sum up these 8 decimal numbers, add the sum result to the perturbation value vx, and then take the modulo 512 to get the rule number x; the calculation formula of the perturbation value vx is as follows:

[0019]

[0020] where, hd i represents the i-th bit in the hash value of the plaintext image;

[0021] Recombine each group of binary keys into a rectangular form of 8×8 as the initial input of the cellular automaton, iterate 200 times according to the rule number B82 / D148, convert the result of the last iteration into a decimal number for every 8 bits, obtaining 8 decimal numbers, sum up these 8 decimal numbers, add the sum result to the perturbation value vy, and then take the modulo 512 to get the rule number y; the calculation formula of the perturbation value vy is as follows:

[0022]

[0023] Convert the obtained rule numbers x and y into binary, and calculate the neighbor distribution situations T of x and y respectively xand T y ;

[0024] If the rule numbers x and y satisfy the balance condition T x ×T y then the balanced rule Bx / Dy is obtained; if not, the cellular automaton with rule number y continues the next iteration until the generated rule number satisfies the balance condition.

[0025] Furthermore, the calculation formula for the neighbor distribution of the rule number is as follows:

[0026]

[0027] where j represents the number of surviving neighbors; a j represents whether j belongs to the set determined by the rule number, that is, the j-th bit in the binary of the rule number; (!) represents the factorial operation.

[0028] Furthermore, step 3 specifically includes:

[0029] Convert the region of interest ROI into a one-dimensional array, decompose the length of the ROI into the product of two numbers r and c with the smallest difference, and reorganize the ROI into an r×c rectangular form;

[0030] Convert the reorganized ROI into binary to obtain eight bit planes B0, B1, B2, …, B7, and regard (B0, B7), (B1, B6), (B2, B5), (B3, B4) as the input of a group of cellular automata. According to the B x1 / D y1 、B x2 / D y2 、B x3 / D y3 、B x4 / D y4 、B x5 / D y5 、B x6 / D y6 、B x7 / D y7 、B x8 / D y8 Iteratively encrypt the eight groups of balanced rules. The specific iterative steps are as follows:

[0031] The first round of iteration: Let Regard (B0, B7), (B1, B6), (B2, B5), (B3, B4) as the input of a group of cellular automata. According to B x1 / D y1 、B x2 / D y2 、B x3 / D y3, B x4 / D y4 Iterate times for encryption to obtain configurations D0, D1, D2, D3, D4, D5, D6, D7;

[0032] Second round of iteration: Let B3 = D3, B2 = D2, B1 = D1, B0 = D0, repeat the iteration of the first round to obtain D'0, D’1, D'2, D'3, D'4, D'5, D'6, D'7;

[0033] Third round of iteration: Let B3 = D’3, B2 = D’2, B1 = D’1, B0 = D'0, according to B x5 / D y5 , B x6 / D y6 , B x7 / D y7 , B x8 / D y8 , repeat the iteration of the first round to obtain D”0, D”1, D”2, D”3, D”4, D”5, D”6, D”7;

[0034] Let Then convert the eight bit planes of D”0, D”1, D”2, D”3, D”4, D”5, D”6, D”7 into a pixel matrix to obtain the encrypted region of interest ROI.

[0035] Furthermore, the method steps of embedding information into the image through histogram shifting technology in steps 4 and 6 include:

[0036] Generate the histogram of the image;

[0037] Find the highest point maxgl and the lowest point mingl in the histogram;

[0038] If the lowest point in the histogram is not zero, add 1 to the pixel values equal to mingl in the image, record the position information of these pixel values and convert them into binary to generate overhead information;

[0039] If maxgl < mingl, add 1 to all pixel values between maxgl and mingl in the image, otherwise subtract 1 from all of them;

[0040] Concatenate the overhead information at the head of the information, and intercept to form the information to be embedded according to the maximum embedding capacity of the image. The part exceeding the maximum embedding capacity is transmitted to the decryption end together with the encrypted image;

[0041] Scan the image with adjusted pixel values line by line. Once a pixel with a pixel value of maxgl is encountered, check the value of the information to be embedded. If it is 0, the pixel value of this pixel remains unchanged; if it is 1, the pixel value of this pixel is changed according to the following formula:

[0042]

[0043] Finally, the image after embedding the information is obtained.

[0044] Compared with the prior art by adopting the above technical solutions, the present invention has the following technical effects:

[0045] 1. The present invention adopts the method of encrypting the region of interest, and only encrypts the key regions in the medical image, reducing the total amount of encryption;

[0046] 2. The present invention uses a bionic cellular automaton to encrypt the ROI region of the medical image. Since the bionic cellular automaton is a completely discrete system, the present invention has superior diffusivity and parallelism, improving the encryption efficiency;

[0047] 3. The present invention does not need to transmit the ROI marking information, and uses the histogram shifting technology to embed the ROI marking information into the ciphertext image, reducing the total amount of transmitted bits. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is the flowchart of the method of the present invention;

[0049] Figure 2 is the flowchart of the processing in the ROI extraction stage of the main process;

[0050] Figure 3 is the original image in the embodiment of the present invention;

[0051] Figure 4 is the histogram of the original image;

[0052] Figure 5 is the encrypted image of an embodiment of the present invention;

[0053] Figure 6 is the histogram of the encrypted image;

[0054] Figure 7 is the histogram of the encrypted region of interest. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention:

[0056] This paper designs a medical image encryption method based on cellular automata and the embedded ROI position. This method can extract the region of interest in the medical image and implement encryption protection for it, and embed the electronic medical record and ROI marking information into the non - interested region and the ciphertext image respectively to further reduce the total transmission volume.

[0057] As Figure 1 shown, a medical image encryption method based on cellular automata and the embedded ROI position includes: dividing the image into blocks, extracting the region of interest and non - interested region in the image according to the block mean threshold and recording the ROI marking information; calculating the update rule of the cellular automata according to the key and the hash value of the plaintext image; reorganizing the region of interest into a rectangle with the smallest difference between the length and width, converting it into a binary bit - plane for iterative encryption; using the histogram shifting technique to embed the electronic medical record or the corresponding diagnostic information into the non - interested region; joining the region of interest and the non - interested region to generate a preliminary ciphertext image, and then embedding the ROI marking information into it to obtain the final ciphertext image.

[0058] Extracting the region of interest and non - interested region and recording the corresponding marking information includes: dividing the image into non - overlapping small blocks and calculating their block means; if the mean of the block is greater than a pre - set threshold, this block is regarded as belonging to the region of interest, otherwise it is regarded as belonging to the non - interested region. In this process, a 0 - 1 marking matrix H is obtained, where 1 indicates that the block corresponding to this position belongs to the region of interest and 0 indicates that the block belongs to the non - interested region. Scanning the marking matrix row by row, if there is no 1 in this row, no operation is performed. If there is 1 in this row, find the first and the last 1, and set all the numbers between the middle columns to 1, that is, regard the corresponding block as the region of interest. Record the row number and the two column numbers and convert them into binary to generate the marking information. Scan all the blocks in raster order. If H(i,j)=1, the block at the corresponding position will be regarded as the region of interest, otherwise it will be regarded as belonging to the non - interested region. Convert the region of interest and the non - interested region into one - dimensional arrays.

[0059] The stage of constructing the balance rule of the cellular automata:

[0060] In this scheme, a 512 - bit binary key is used and it is divided into 8 groups to generate 8 groups of balance rules. Since the generation steps of the rule numbers are the same, the following takes the first group as an example to introduce the generation of the first group of balance rules in detail. The generation of the remaining rule numbers is the same as this step, only the data is different.

[0061] Use the MD5 algorithm to calculate the 128 - bit hash value of the plaintext image, and then convert each eight - bit of it into a decimal number to get hd1, hd2,..., hd 16 。

[0062] Calculate the perturbation values vx and vy according to the following formula for establishing the association between the encryption method and the plaintext image:

[0063]

[0064] Convert the first part k1, k2,..., k of the key 64 to decimal numbers R1, R2,..., R8. The calculation of rule number x1 is as follows:

[0065]

[0066] Convert the rule number x1 to a binary number and calculate the number T of its neighbor distribution x .

[0067]

[0068] where j represents the number of surviving neighbors; a j indicates whether j representing the survival number belongs to the set determined by the rule number.

[0069] Recombine the key k1, k2,..., k 64 into an 8×8 rectangle and use it as the input of the cellular automaton. Iterate 200 times according to rule B82 / D148 to generate ek1, ek2,..., ek 64 . Then, similar to generating the rule number x1, convert ek1, ek2,..., ek 64 every 8 bits to decimal to obtain R9, R 10 ,..., R 16 , and calculate the rule number y1:

[0070]

[0071] Convert the rule number y1 to a binary number and calculate the number T of neighbor distribution using formula (3) y .

[0072] Conduct a balance test: If the rule numbers x1 and y1 satisfy the balance condition T x = T y then construct the balance rule Bx1 / Dy1; if not, the cellular automaton with rule number y1 will be iterated one more time to recalculate the rule number y1 until the balance condition is satisfied.

[0073] In the region of interest extraction stage, as Figure 2 shown, it includes the following steps:

[0074] As an implementation, the plaintext image is a standard test grayscale image Brain with a size of 256×256. Divide it into multiple non-overlapping blocks of 16×16. Obtain the block image set Pblock = [P1, P2,..., P mn , Then reorganize P block into m × n.

[0075] Calculate the mean value of each block to generate MP = [mp 11 , mp 12 ,..., mp mn .

[0076] Define a all-zero matrix H of m × n and a block mean threshold ts = 20. Scan MP row by row. If mp ij ≥ ts, then the value at the position corresponding to this element in H becomes 1.

[0077] Scan H row by row. For each row i, we check whether there is a 1 in this row. If not, skip this row; if there is, find the position of the first value of 1, denoted as row H(i, col1) = 1, and find the last position of 1, denoted as H(i, col2) = 1. Record the current i, col1, and col2 into the marking information, denoted as RH = [RH; i, col1, col2].

[0078] Scan P in the order of the diaphragm block . If H(i, j) = 1, then the corresponding P block will be regarded as the region of interest, otherwise it will be regarded as belonging to the non - region of interest. Convert the region of interest and the non - region of interest into one - dimensional arrays with lengths lroi and lroni respectively.

[0079] Convert the marking information RH into a binary sequence BRH with a fixed length. For each triple (i, col1, col2), the row number i requires bits to represent, and col1 and col2 require bits.

[0080] Encryption stage:

[0081] Split the length lroi of the ROI into the product of r and c, ensuring that |r - c| is minimized. Then rearrange the ROI into r × c. For example, if the ROI in this implementation example is 1 × 26112, it can be reorganized into 136 × 192.

[0082] Assume that each pixel value in the ROI consists of 8 - bit binary numbers. Convert the ROI from a pixel - value matrix to 8 binary bit - planes B7, B6, B5, B4, B3, B2, B1, B0. The size of each bit - plane is r × c.

[0083] Perform the first - round encryption with a bionic cellular automaton: Let Regard (B0, B7), (B1, B6), (B2, B5), (B3, B4) as the input of a group of cellular automata. According to the calculated rule number B x1 / D y1 、B x2 / D y2 、B x3 / D y3 、B x4 / D y4 Iterate times for encryption to obtain the final configurations D0, D1, D2, D3, D4, D5, D6, D7.

[0084] Second-round encryption: Let B3 = D3, B2 = D2, B1 = D1, B0 = D0. Repeat the iterative encryption of the first round to obtain D'0, D’1, D'2, D'3, D'4, D'5, D'6, D'7.

[0085] Third-round encryption: Let B3 = D’3, B2 = D’2, B1 = D’1, B0 = D'0. Use the new rule numbers B x5 / D y5 、B x6 / D y6 、B x7 / D y7 、B x8 / D y8 Repeat the iterative encryption of the first round to obtain D”0, D”1, D”2, D”3, D”4, D”5, D”6, D”7.

[0086] Let Then convert these 8 bit-planes into a pixel matrix to obtain the encrypted ROI: enROI.

[0087] Embedding stage:

[0088] This method uses the histogram shifting technique to embed the electronic medical record and ROI marker information into RONI and the ciphertext image respectively. Since there must be zeros in the histogram of the RONI region, the following takes the embedding of marker information as an example to describe the embedding steps in detail: Figure 1 There must be zeros in the histogram of the RONI region. Therefore, the following takes the embedding of marker information as an example to describe the embedding steps in detail:

[0089] Generate the histogram of the ciphertext image, and find the highest and lowest points in the histogram, maxgl and mingl.

[0090] If the lowest point mingl in the histogram is not zero, add 1 to the pixel values equal to mingl (subtract 1 if the pixel value is 255), record the positions of these pixels and convert them into binary to generate the overhead information overinfo.

[0091] Change the pixel values between maxgl and mingl. If maxgl < mingl, add 1 to the pixel value; otherwise, subtract 1.

[0092] Construct the information to be embedded eInfo. If there is overhead information, insert it at the head of the input information, that is

[0093] Intercept the information to be embedded according to the maximum embedding capacity, that is, eInfo = eInfo(1:H(maxgl)). The part gInfo = eInfo(H(maxgl)+1:le) that exceeds the maximum embedding capacity is transmitted to the decryption end together with the ciphertext image.

[0094] Scan the input image line by line. Once a pixel with a value of maxgl is encountered, check the value of the information to be embedded. If it is 0, do not change the current pixel value; if it is 1, change the pixel value to:

[0095]

[0096] Output the ciphertext image after embedding the information.

[0097] This specific implementation example uses MATLAB R2016a software for simulation. The plaintext image selects the standard test grayscale image Brain with a size of 256×256, as Figure 3 shown; Figure 4 shows the histogram of the plaintext image of an implementation manner of the present invention; Figure 5 shows the ciphertext image of an implementation manner of the present invention, Figure 6 shows the histogram of the ciphertext image of an implementation manner of the present invention; Figure 7 shows the histogram after encrypting the region of interest of an implementation manner of the present invention.

[0098] Histogram analysis: The color histogram reflects the gray level distribution of pixels. Generally speaking, there is a specific distribution rule for the histogram of an image, which may leak some information. Therefore, the histogram of the encrypted image should be close to a uniform distribution to hide this information. By comparing Figure 4 and Figure 6 , it can be seen that the histogram of the ciphertext image is different from that of the plaintext image. As Figure 7 shown, the histogram of the region of interest in the encrypted image is uniform and random, which can effectively resist statistical attacks.

[0099] Correlation analysis: The correlation of adjacent pixels reflects the degree of correlation of pixel values at adjacent positions in the horizontal, vertical, and diagonal directions of the image. For an excellent encryption scheme, the correlation of the ciphertext image should be as close as possible to the theoretical value of 0. In the present invention, 20,000 pairs of adjacent pixels are randomly selected from each direction of the plaintext image and the ciphertext image for correlation testing. The correlation of the encrypted image is quantitatively analyzed according to Equation (6).

[0100]

[0101] Among them, x and y represent the pixel values of two adjacent pixels.

[0102] Table 1 shows the comparison of the correlation coefficients of the plaintext image and the full image encrypted by the present invention in the horizontal direction, vertical direction, and diagonal direction. It can be seen that the correlation of adjacent pixels in the plaintext image is close to 1 in all three directions, indicating a high correlation of the original image. And these values in the ciphertext image are close to 0, indicating a low correlation of the pixels in the ciphertext image. The present invention can effectively resist statistical attacks based on pixel correlation.

[0103] Table 1 Comparison between the plaintext image and the full image encrypted by the present invention

[0104]

[0105] In summary, a method for encrypting regions of interest in medical images provided by the embodiments of the present invention has good encryption effect, is robust against various attacks, and has high encryption efficiency.

[0106] The above is only the specific implementation manner in the present invention, but the protection scope of the present invention is not limited thereto. Any lossless embedding technology that can be understood and conceived by those familiar with the technology within the technical scope disclosed by the present invention should be covered within the scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A medical image encryption method based on cellular automata and the embedded ROI position, characterized in that, The method includes: Step 1: Divide the plaintext image into blocks, extract the region of interest (ROI) and the region of non-interest (RONI) in the original image based on the mean value of each block, and record the corresponding ROI marking information; Step 2: Calculate eight groups of balanced rules of the cellular automaton according to the key and the hash value of the plaintext image; Step 3: Recombine the region of interest into a rectangle with the smallest difference between the length and width, and use the cellular automaton for iterative encryption; Step 4: Embed the binary-form diagnostic information into the region of non-interest (RONI) through the histogram shifting technique; Step 5: Combine the encrypted region of interest (ROI) and the region of non-interest (RONI) with the embedded information to form a preliminary ciphertext image C1; Step 6: Embed the ROI marking information in Step 1 into the preliminary ciphertext image C1 through the histogram shifting technique to obtain the final encrypted image; Step 3 specifically includes: Convert the region of interest (ROI) into a one-dimensional array, decompose the length of the ROI into the product of two numbers r and c with the smallest difference, and recombine the ROI into an r×c rectangular form; Convert the recombined ROI into binary to obtain eight bit planes B0, B1, B2, …, B7. Consider (B0, B7), (B1, B6), (B2, B5), (B3, B4) as the inputs of a group of cellular automata, and according to B obtained in step 2 x1 / D y1 、B x2 / D y2 、B x3 / D y3 、B x4 / D y4 、B x5 / D y5 、B x6 / D y6 、B x7 / D y7 、B x8 / D y8 Perform iterative encryption on eight groups of balance rules. The specific iterative steps are as follows: The first round of iteration: Let Regard (B0, B7), (B1, B6), (B2, B5), (B3, B4) as the input of a set of cellular automata. According to B x1 / D y1 , B x2 / D y2 , B x3 / D y3 , B x4 / D y4 Iterate times for encryption to obtain configurations D0, D1, D2, D3, D4, D5, D6, D7; Second iteration: Let B3 = D3, B2 = D2, B1 = D1, B0 = D0, repeat the iteration of the first round to obtain D′0, D1′, D′2, D′3, D′4, D′5, D′6, D′7; Third iteration: Let ” B3 = D3, B2 = D2, B1 = D1′, B0 = D′0, according to B x5 / D y5 、B x6 / D y6 、B x7 / D y7 、B x8 / D y8 , repeat the iteration of the first round to obtain D″0″, D1″, D″2″, D″3″, D″4″, D″5″, D″6″, D″7″; Let Then convert the eight bit planes D″0″, D1”, D″2″, D″3″, D″4″, D″5″, D″6″, D″7″ into a pixel matrix to obtain the encrypted region of interest ROI.

2. A medical image encryption method based on cellular automata and embedding ROI positions according to claim 1, characterized in that Step 1 specifically includes: Divide the plaintext image into blocks, calculate the pixel mean value in each block respectively, and set up a matrix H. If the pixel mean value in the block located at the i-th row and j-th column of the plaintext image is greater than the preset block mean threshold, then the element H(i,j) at the i-th row and j-th column in H is 1, otherwise H(i,j) = 0; Scan the matrix H row by row. If there is no element with a value of 1 in a certain row, do not perform any processing and proceed to scan the next row; if there is an element with a value of 1, then set all the elements between the first element with a value of 1 and the last element with a value of 1 in this row to 1, and convert the row number and the column numbers corresponding to the first element with a value of 1 and the last element with a value of 1 into binary to generate a triple; after completing the scan, obtain the marking matrix H′ and the marking information composed of all triples; If H′(i,j) = 1, the corresponding block belongs to the region of interest (ROI), otherwise it belongs to the region of non-interest (RONI).

3. A medical image encryption method based on cellular automata and the embedded ROI position according to claim 1, characterized in that The specific calculation method of each group of balanced rules in Step 2 includes: Divide the 512-bit binary key into eight groups, with each group containing 64-bit binary numbers; convert every eight bits in the 64-bit binary numbers of each group into a decimal number, obtaining 8 decimal numbers within the range (0, 255). Sum up these 8 decimal numbers, add the sum result to the perturbation value vx, and take the modulus of 512 to obtain the rule number x; the calculation formula of the perturbation value vx is as follows: Among them, hd i represents the i-th bit in the hash value of the plaintext image; Recombine each group of binary keys into an 8×8 rectangular form as the initial input of the cellular automaton, iterate 200 times according to the rule number B82 / D148, convert the result of the last iteration into a decimal number every 8 bits, obtaining 8 decimal numbers. Sum up these 8 decimal numbers, add the sum result to the perturbation value vy, and take the modulus of 512 to obtain the rule number y; the calculation formula of the perturbation value vy is as follows: Convert the obtained rule numbers x and y into binary, and calculate the neighbor distribution situations T of x and y respectively x and T y ; If the rule numbers x and y satisfy the balance condition T x = T y then the balanced rule Bx / Dy is obtained; if not, the cellular automaton with rule number y continues the next iteration until the generated rule number satisfies the balance condition.

4. A medical image encryption method based on cellular automata and the embedded ROI position according to claim 3, characterized in that, The calculation formula of the neighbor distribution of the rule number is as follows: where j represents the number of surviving neighbors; a j represents whether j belongs to the set determined by the rule number, that is, the j-th bit in the binary of the rule number; (·)! represents the factorial operation.

5. A medical image encryption method based on cellular automata and the position of embedded ROI according to claim 1, characterized in that The method steps of embedding information into the image through the histogram shifting technique in Step 4 and 6 include: Generate the histogram of the image; Find the highest point maxgl and the lowest point mingl in the histogram; If the lowest point in the histogram is not zero, add 1 to the pixel values equal to mingl in the image, record the position information of these pixel values and convert them to binary to generate overhead information; If maxgl < mingl, add 1 to all pixel values between maxgl and mingl in the image, otherwise subtract 1 from all of them; Concatenate the overhead information to the head of the information and intercept it according to the maximum embedding capacity of the image to form the information to be embedded. The part exceeding the maximum embedding capacity is transmitted to the decryption end together with the encrypted image; Scan the image with adjusted pixel values line by line. Once a pixel with a pixel value of maxgl is encountered, check the value of the information to be embedded. If it is 0, the pixel value of this pixel remains unchanged; if it is 1, the pixel value of this pixel changes according to the following formula: Finally, obtain the image after embedding the information.

Images