UAV remote sensing image protection method based on two-dimensional chaotic mapping for palmprint verification

Abstract

The present invention discloses a method for protecting UAV remote sensing images based on a two-dimensional chaotic map for palm print verification. The new two-dimensional chaotic map adopted has the advantages of being sufficiently sensitive to the key, difficult to be cracked, and capable of generating sufficiently random sequences. The adopted step-type scanning algorithm is divided into three parts. The first part corresponds to the lower left part of the plaintext UAV remote sensing image, the second part corresponds to the diagonal part of the plaintext UAV remote sensing image, and the third part corresponds to the upper right part of the plaintext UAV remote sensing image. Both the first and third parts cross the boundary and have extremely weak periodicity. The scanning route adopts a step-type, complex scanning route, and the scanning scrambling effect is very good. Compared with traditional fission diffusion, the adopted three-dimensional fission diffusion can encrypt images of any size and encrypt color images. There are more diffusion directions, namely row direction, column direction, and vertical direction, which makes the diffusion effect better.

Description

Technical Field

[0001] The present invention relates to the technical field of information security, and in particular to a method for protecting unmanned aerial vehicle (UAV) remote sensing images based on two-dimensional chaotic mapping for palmprint verification. Background Art

[0002] With the rapid development of internet technology, data storage and transmission security face significant challenges, especially in applications such as drone remote sensing imagery, where real-time transmission of images containing sensitive geographic or environmental information is crucial. The secure transmission of these images requires the use of efficient and secure encryption technologies to prevent illegal access, theft, or unauthorized use, thereby avoiding serious security issues and financial losses.

[0003] Due to the characteristics of digital images, such as large data capacity and strong correlations between pixels, traditional encryption algorithms such as AES and DES are no longer applicable. Chaotic systems, as nonlinear dynamic systems, are ideal for digital image encryption due to their sensitivity to initial values ​​and parameters, ergodicity, aperiodicity, and randomness, as well as the complex pseudo-random sequence structures generated by chaotic mapping. In particular, the research and application of hyperchaotic systems demonstrates greater randomness and unpredictability than conventional chaos. However, existing chaos-based image encryption methods still suffer from the following issues: 1. The periodicity of chaotic mapping results in insufficient randomness in the generated sequence; 2. Scanning scrambling is ineffective; and 3. Fission-diffusion can only encrypt square images. Summary of the Invention

[0004] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a UAV remote sensing image protection method based on two-dimensional chaotic mapping for palmprint verification. The method can not only solve the problems of periodicity and insufficient randomness of the generated sequence, but also has the advantages of good scanning scrambling effect and the ability to encrypt images of any size.

[0005] To achieve the above objectives, the technical solutions provided by the present invention are:

[0006] The UAV remote sensing image protection method based on two-dimensional chaotic mapping for palmprint verification includes:

[0007] S1, extract the characteristics of the palmprint image of the ciphertext recipient as the key;

[0008] S2, obtaining the initial parameters required for the two-dimensional chaotic mapping through the key;

[0009] S3, combining the initial parameters, generating a chaotic sequence through a two-dimensional chaotic map, and obtaining a key matrix based on the chaotic sequence;

[0010] S4, performing scanning scrambling processing on the UAV remote sensing image to obtain a scrambled UAV remote sensing image;

[0011] S5, performing three-dimensional fission diffusion processing on the scrambled UAV remote sensing image in combination with the key matrix to obtain an encrypted image;

[0012] S6. When the ciphertext receiver needs to obtain the plaintext medical drone remote sensing image, it regenerates the key matrix corresponding to its key;

[0013] S7, combining the key matrix generated in step S6, performing three-dimensional fission inverse diffusion processing on the encrypted image to obtain a scrambled image;

[0014] S8. Perform inverse scrambling processing on the scrambled image obtained in step S7, thereby obtaining the plaintext UAV remote sensing image of the ciphertext receiver.

[0015] Furthermore, the process of step S3 includes:

[0016] Step A1, initialize variables and set parameter values: formula (x0, y0); number of iterations i = 1; N0 is the set value, obtained according to the key; given the values ​​of parameters a and b in the two-dimensional chaos formula, obtain the length M and width N of the original image;

[0017] Step A2: Perform an iterative loop Obtained The size of the components is Array sequence x, where L = M*N:

[0018] D1. Update the value of x(i):

[0019] D2, i=i+1;

[0020] Step A3: Interception interception Connect the two random sequences as a new sequence x;

[0021] Step A4: Calculate x: x = mod (floor (x * 10^6), 256);

[0022] Step A5: Calculate y: y = mod(floor(x*10^6), 256);

[0023] Step A6: Calculate z: z = mod (floor (x * 10^6), 256);

[0024] Step A7: Use the Reshape function to transform the chaotic sequences x, y, and z into key matrices X 2D 、Y 2D , Z 2D ;

[0025] L is the length of the two-dimensional vector after the plaintext UAV remote sensing image is converted into a two-dimensional vector.

[0026] Furthermore, in the process of scanning and scrambling the plaintext UAV remote sensing image, the scanning is divided into three parts. The first part corresponds to the lower left part of the plaintext UAV remote sensing image, the second part corresponds to the diagonal part of the plaintext UAV remote sensing image, and the third part corresponds to the upper right part of the plaintext UAV remote sensing image.

[0027] Furthermore, the scan of the lower left part yields the array Line1:

[0028] Line1[2(k-1)^2+j]=img[m-2(k-1)+floor(j / 2),floor((j+1) / 2)]

[0029] in, and j=1,2,...,4(s-1)+1;

[0030] Line1[2(k-1)^2+j] represents the value of the 2(k-1)^2+jth pixel in Line1, and img[m-2(k-1)+floor(j / 2),floor((j+1) / 2)] represents the pixel value of the m-2(k-1)+floor(j / 2)th row and floor((j+1) / 2)th column in the image;

[0031] Scanning the diagonal lines yields the array Line2:

[0032] Line2[s]=img[s,s]

[0033] Where s = 1, 2, ..., n;

[0034] Scanning the upper right part yields array Line3:

[0035]

[0036] where k = (n-2) / 2+1, (n-2) / 2, ..., 1 and j = 1, 2, ..., 4(k-1)+1)

[0037] Indicates the first The value of the pixel at each point, and img[floor((j+1) / 2),2x((n-2) / 2+1+1-k)+floor(j / 2)] represents the pixel value at the floor((j+1) / 2)th row and the 2x((n-2) / 2+1+1-k)+floor(j / 2)th column in the image.

[0038] Furthermore, the scanned image is scrambled, and the process includes:

[0039] Concatenate Line1, Line2, and Line3 to get Line:

[0040] Line = Line1||Line2||Line3;

[0041] The symbol || means that the three arrays Line1, Line2, and Line3 are horizontally concatenated in order into a single array Line, that is, an array matrix with a length of 1 and a width of n*n;

[0042] Use the Reshape formula to transform the scanned array matrix with a length of 1 and a width of n*n into a matrix with a length of n and a width of n: Line = Reshape(Line, n, n), and the scrambled UAV remote sensing image is obtained.

[0043] Furthermore, the process of three-dimensional fission diffusion includes a positive sequence stage and a reverse sequence stage;

[0044] The image P is divided into three layers: R, G, and B;

[0045] In the forward order phase, the encryption process starts from the upper left corner of the plaintext image and spreads to the lower right corner. For each pixel other than layers (1, 1) and (M, N), the pixel is affected by the pixels to the left and above in the same layer and other layers. After the forward order phase, the image P is converted to image D.

[0046] In the inverse stage, it runs in the opposite order, starting from the lower right corner and diffusing to the upper left corner. The pixels are affected by the pixels on the right and below in the same layer and other layers. After the inverse stage, image D is converted to image E.

[0047] Furthermore, the positive sequence phase includes the following:

[0048] B1. Encrypt the pixel at (1,1) in each layer of image P using the following formula:

[0049] D(1,1,1)=mod(P(1,1,1)+X 2D (1,1),256)

[0050] D(1,1,2)=mod(P(1,1,2)+Y 2D (1,1),256)

[0051] D(1,1,3)=mod(P(1,1,3)+Z 2D (1,1),256)

[0052] B2. Encrypt the pixels of the first row and first column of each layer of the image P using the following two formulas:

[0053] D(i,1,1)=mod(P(i,1,1)+X 2D (i,1)+D(i-1,1,1)+D(i-1,1,3),256)

[0054] D(i,1,2)=mod(P(i,1,2)+Y 2D (i,1)+D(i-1,1,2)+D(i-1,1,1),256)

[0055] D(i,1,3)=mod(P(i,1,3)+Z 2D (i,1)+D(i-1,1,3)+D(i-1,1,2),256)

[0056] D(1,j,1)=mod(P(1,j,1)+X 2D (1,j)+D(1,j-1,1)+D(1,j-1,3),256)

[0057] D(1,j,2)=mod(P(1,j,2)+Y 2D (1,j)+D(1,j-1,2)+D(1,j-1,1),256)

[0058] D(1,j,3)=mod(P(1,j,3)+Z 2D (1,j)+D(1,j-1,3)+D(1,j-1,2),256)

[0059] B3. Use the following formula to encrypt the remaining pixels in each layer of image P, and finally obtain image D:

[0060] D(i,j,1)=mod(P(i,j,1)+X 2D (i,j)+D(i,j-1,3)+D(i-1,j,3),256)

[0061] D(i,j,2)=mod(P(i,j,2)+Y 2D (i,j)+D(i,j-1,1)+D(i-1,j,1),256)

[0062] D(i,j,3)=mod(P(i,j,3)+Z 2D (i,j)+D(i,j-1,2)+D(i-1,j,2),256)

[0063] The reverse phase includes the following:

[0064] C1. Encrypt the (M, N) and (1, 1) pixels of image D at each layer using the following formula:

[0065] E(M,N,1)=mod(D(1,1,1)+X 2D (M,N),256)

[0066] E(M,N,2)=mod(D(1,1,2)+Y 2D (M,N),256)

[0067] E(M,N,3)=mod(D(1,1,3)+Z 2D (M,N),256)

[0068] C2. For the pixels in the Mth and Nth rows of each layer of image D, encrypt them using the following two formulas:

[0069] E(i,N,1)=mod(D(i,N,1)+X 2D (i,N)+E(i+1,N,1)+E(i+1,N,2),256)

[0070] E(i,N,2)=mod(D(i,N,2)+Y 2D (i,N)+E(i+1,N,2)+E(i+1,N,3),256)

[0071] E(i,N,3)=mod(D(i,N,3)+Z 2D (i,N)+E(i+1,N,3)+E(i+1,N,1),256)

[0072] E(M,j,1)=mod(D(M,j,1)+X 2D (M,j)+D(M,j+1,1)+D(M,j+1,2),256)

[0073] E(M,j,2)=mod(D(M,j,2)+Y 2D (M,j)+D(M,j+1,2)+D(M,j+1,3),256)

[0074] E(M,j,3)=mod(D(M,j,3)+Z 2D (M,j)+D(M,j+1,3)+D(M,j+1,1),256)

[0075] C3. For the remaining pixels in image D, encrypt using the following formula to obtain the final encrypted image E:

[0076] E(i,j,1)=mod(D(i,j,1)+X 2D(i,j)+E(i,j+1,3)+E(i+1,j,2),256)

[0077] E(i,j,2)=mod(D(i,j,2)+Y 2D (i,j)+E(i,j+1,3)+E(i+1,j,3),256)

[0078] E(i,j,3)=mod(D(i,j,3)+Z 2D (i,j)+E(i,j+1,2)+E(i+1,j,1),256)

[0079] Among them, X 2D 、Y 2D , Z 2D is the corresponding key matrix.

[0080] Compared with the existing technology, the principles and advantages of this solution are as follows:

[0081] 1. Traditional two-dimensional mapping has problems such as being insufficiently sensitive to keys, being easily cracked, being periodic, and generating insufficient randomness in sequences. The new two-dimensional chaotic mapping used in this technical solution has the advantages of being sufficiently sensitive to keys, being difficult to crack, and being able to generate sufficiently random sequences.

[0082] 2. Traditional scanning algorithms all scan within a matrix, which leads to periodicity issues. Furthermore, the scanning route is simple, resulting in poor scan scrambling effects. The staircase scanning algorithm used in this technical solution is divided into three parts: the first part corresponds to the lower left portion of the plaintext UAV remote sensing image, the second part corresponds to the diagonal portion of the plaintext UAV remote sensing image, and the third part corresponds to the upper right portion of the plaintext UAV remote sensing image. Both the first and third parts cross the boundary, resulting in extremely weak periodicity. The staircase scanning route is complex, yet the scan scrambling effect is excellent.

[0083] 3. Traditional fission diffusion cannot encrypt images of any size, only square images. Furthermore, the diffusion direction is limited. However, the algorithm in this technical solution uses three-dimensional fission diffusion, which, compared to traditional fission diffusion, can encrypt images of any size and color images. Diffusion has more directions: row, column, and vertical, resulting in a more effective diffusion effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the services required for use in the embodiments or the prior art descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0085] Figure 1 This is a principle flow chart of the UAV remote sensing image protection method based on two-dimensional chaotic mapping with palmprint feature verification according to the present invention;

[0086] Figure 2 This is a schematic diagram of the principle of the scanning algorithm in the UAV remote sensing image protection method based on two-dimensional chaotic mapping with palmprint feature verification of the present invention;

[0087] Figure 3 This is a graph showing the three-dimensional Lyapunov test results of the two-dimensional chaos of the present invention;

[0088] Figure 4 、 Figure 5 This is a comparison chart of the two-dimensional Lyapunov test results;

[0089] Figure 6 is the bifurcation diagram of variable a;

[0090] Figure 7 is the bifurcation diagram of variable b;

[0091] Figure 8 For various existing scanning methods;

[0092] Figure 9 is the histogram of the UAV remote sensing image before scrambling;

[0093] Figure 10 is the histogram of the UAV remote sensing image after scrambling;

[0094] Figure 11 is the image after diffusion;

[0095] Figure 12 is the histogram of the image after diffusion;

[0096] Figure 13 Schematic diagram of the positive sequence stage of the three-dimensional fission-diffusion algorithm. DETAILED DESCRIPTION

[0097] The present invention will be further described below in conjunction with specific embodiments:

[0098] like Figure 1 As shown, the UAV remote sensing image protection method based on two-dimensional chaotic mapping of palmprint verification described in this embodiment includes the following steps:

[0099] S1, extract the characteristics of the palmprint image of the ciphertext recipient as the key;

[0100] S2, obtaining the initial parameters required for the two-dimensional chaotic mapping through the key;

[0101] S3, combining the initial parameters, generating a chaotic sequence through a two-dimensional chaotic map, and obtaining a key matrix based on the chaotic sequence;

[0102] The specific process of this step includes:

[0103] Step A1, initialize variables and set parameter values: formula (x0, y0); number of iterations i = 1; N0 is the set value, obtained according to the key; given the values ​​of parameters a and b in the two-dimensional chaos formula, obtain the length M and width N of the original image;

[0104] Step A2: Perform an iterative loop Obtained The size of the components is Array sequence x, where L = M*N:

[0105] D1. Update the value of x(i):

[0106] D2, i=i+1;

[0107] Step A3: Interception interception Connect the two random sequences as a new sequence x;

[0108] Step A4: Calculate x: x = mod (floor (x * 10^6), 256);

[0109] Step A5: Calculate y: y = mod(floor(x*10^6), 256);

[0110] Step A6: Calculate z: z = mod (floor (x * 10^6), 256);

[0111] Step A7: Use the Reshape function to transform the chaotic sequences x, y, and z into key matrices X 2D 、Y 2D , Z 2D ;

[0112] L is the length of the two-dimensional vector after the plaintext UAV remote sensing image is converted into a two-dimensional vector.

[0113] S4, performing scanning scrambling processing on the UAV remote sensing image to obtain a scrambled UAV remote sensing image;

[0114] In the process of scanning and scrambling plaintext UAV remote sensing images, Figure 2 As shown, the scanning is divided into three parts. The first part corresponds to the lower left part of the plaintext UAV remote sensing image, the second part corresponds to the diagonal part of the plaintext UAV remote sensing image, and the third part corresponds to the upper right part of the plaintext UAV remote sensing image.

[0115] Scanning the lower left part yields array Line1:

[0116] Line1[2(k-1)^2+j]=img[m-2(k-1)+floor(j / 2),floor((j+1) / 2)]

[0117] in, and j=1,2,...,4(s-1)+1;

[0118] Line1[2(k-1)^2+j] represents the value of the 2(k-1)^2+jth pixel in Line1, and img[m-2(k-1)+floor(j / 2),floor((j+1) / 2)] represents the pixel value of the m-2(k-1)+floor(j / 2)th row and floor((j+1) / 2)th column in the image;

[0119] Scanning the diagonal lines yields the array Line2:

[0120] Line2[s]=img[s,s]

[0121] Where s = 1, 2, ..., n;

[0122] Scanning the upper right part yields array Line3:

[0123]

[0124] where k = (n-2) / 2+1, (n-2) / 2, ..., 1 and j = 1, 2, ..., 4(k-1)+1)

[0125] Indicates the first The value of the pixel at each point, and img[floor((j+1) / 2),2x((n-2) / 2+1+1-k)+floor(j / 2)] represents the pixel value at the floor((j+1) / 2)th row and the 2x((n-2) / 2+1+1-k)+floor(j / 2)th column in the image.

[0126] The scanned image is scrambled. The process includes:

[0127] Concatenate Line1, Line2, and Line3 to get Line:

[0128] Line = Line1||Line2||Line3;

[0129] The symbol || means that the three arrays Line1, Line2, and Line3 are horizontally concatenated in order into a single array Line, that is, an array matrix with a length of 1 and a width of n*n;

[0130] Use the Reshpe formula to transform the scanned array matrix with a length of 1 and a width of n*n into a matrix with a length of n and a width of n: Line = Reshape(Line, n, n), and the scrambled UAV remote sensing image is obtained.

[0131] S5, performing three-dimensional fission diffusion processing on the scrambled UAV remote sensing image in combination with the key matrix to obtain an encrypted image;

[0132] The process of three-dimensional fission diffusion includes the positive sequence stage and the reverse sequence stage;

[0133] The image P is divided into three layers: R, G, and B;

[0134] In the forward order phase, the encryption process starts from the upper left corner of the plaintext image and spreads to the lower right corner. For each pixel other than layers (1, 1) and (M, N), the pixel is affected by the pixels to the left and above in the same layer and other layers. After the forward order phase, the image P is converted to image D.

[0135] The positive sequence phase includes the following:

[0136] B1. Encrypt the pixel at (1,1) in each layer of image P using the following formula:

[0137] D(1,1,1)=mod(P(1,1,1)+X 2D (1,1),256)

[0138] D(1,1,2)=mod(P(1,1,2)+Y 2D (1,1),256)

[0139] D(1,1,3)=mod(P(1,1,3)+Z 2D (1,1),256)

[0140] B2. Encrypt the pixels of the first row and first column of each layer of the image P using the following two formulas:

[0141] D(i,1,1)=mod(P(i,1,1)+X 2D(i,1)+D(i-1,1,1)+D(i-1,1,3),256)

[0142] D(i,1,2)=mod(P(i,1,2)+Y 2D (i,1)+D(i-1,1,2)+D(i-1,1,1),256)

[0143] D(i,1,3)=mod(P(i,1,3)+Z 2D (i,1)+D(i-1,1,3)+D(i-1,1,2),256)

[0144] D(1,j,1)=mod(P(1,j,1)+X 2D (1,j)+D(1,j-1,1)+D(1,j-1,3),256)

[0145] D(1,j,2)=mod(P(1,j,2)+Y 2D (1,j)+D(1,j-1,2)+D(1,j-1,1),256)

[0146] D(1,j,3)=mod(P(1,j,3)+Z 2D (1,j)+D(1,j-1,3)+D(1,j-1,2),256)

[0147] B3. Use the following formula to encrypt the remaining pixels in each layer of image P, and finally obtain image D:

[0148] D(i,j,1)=mod(P(i,j,1)+X 2D (i,j)+D(i,j-1,3)+D(i-1,j,3),256)

[0149] D(i,j,2)=mod(P(i,j,2)+Y 2D (i,j)+D(i,j-1,1)+D(i-1,j,1),256)

[0150] D(i,j,3)=mod(P(i,j,3)+Z 2D (i,j)+D(i,j-1,2)+D(i-1,j,2),256)

[0151] In the inverse stage, it runs in the opposite order, starting from the lower right corner and diffusing to the upper left corner. The pixels are affected by the pixels on the right and below in the same layer and other layers. After the inverse stage, image D is converted to image E.

[0152] The reverse phase includes the following:

[0153] C1. Encrypt the (M, N) and (1, 1) pixels of image D at each layer using the following formula:

[0154] E(M,N,1)=mod(D(1,1,1)+X 2D (M,N),256)

[0155] E(M,N,2)=mod(D(1,1,2)+Y 2D (M,N),256)

[0156] E(M,N,3)=mod(D(1,1,3)+Z 2D (M,N),256)

[0157] C2. For the pixels in the Mth and Nth rows of each layer of image D, encrypt them using the following two formulas:

[0158] E(i,N,1)=mod(D(i,N,1)+X 2D (i,N)+E(i+1,N,1)+E(i+1,N,2),256)

[0159] E(i,N,2)=mod(D(i,N,2)+Y 2D (i,N)+E(i+1,N,2)+E(i+1,N,3),256)

[0160] E(i,N,3)=mod(D(i,N,3)+Z 2D (i,N)+E(i+1,N,3)+E(i+1,N,1),256)

[0161] E(M,j,1)=mod(D(M,j,1)+X 2D (M,j)+D(M,j+1,1)+D(M,j+1,2),256)

[0162] E(M,j,2)=mod(D(M,j,2)+Y 2D (M,j)+D(M,j+1,2)+D(M,j+1,3),256)

[0163] E(M,j,3)=mod(D(M,j,3)+Z 2D (M,j)+D(M,j+1,3)+D(M,j+1,1),256)

[0164] C3. For the remaining pixels in image D, encrypt using the following formula to obtain the final encrypted image E:

[0165] E(i,j,1)=mod(D(i,j,1)+X 2D(i,j)+E(i,j+1,3)+E(i+1,j,2),256)

[0166] E(i,j,2)=mod(D(i,j,2)+Y 2D (i,j)+E(i,j+1,3)+E(i+1,j,3),256)

[0167] E(i,j,3)=mod(D(i,j,3)+Z 2D (i,j)+E(i,j+1,2)+E(i+1,j,1),256).

[0168] S6. When the ciphertext receiver needs to obtain the plaintext medical drone remote sensing image, it regenerates the chaotic sequence and key matrix corresponding to its key;

[0169] S7: Combine the key matrix generated in step S6 and perform three-dimensional fission diffusion processing on the encrypted image to obtain a scrambled image.

[0170] S8. Perform inverse scrambling processing on the scrambled image obtained in step S7, thereby obtaining the plaintext UAV remote sensing image of the ciphertext receiver.

[0171] In order to demonstrate the effectiveness and superiority of the UAV remote sensing image protection method of the present invention, the following test experiments were conducted:

[0172] 1. Lyapunov test: Figure 3 The three-dimensional Lyapunov exponent image of the two-dimensional chaotic mapping proposed in this invention is: Figure 4 and Figure 5 This figure compares the Lyapunov exponents of the proposed 2D chaotic map with those of other 2D chaotic maps. Higher Lyapunov exponent values ​​indicate weaker periodicity and greater difficulty in cracking. Compared to the 2D chaotic map and the 2D chaotic map, the control parameters a and b of the proposed chaotic map exhibit significantly higher Lyapunov values ​​when the values ​​are between 0 and 25, respectively.

[0173] 2. Bifurcation diagram test: Figure 6 and Figure 7 This is the bifurcation diagram of the two-dimensional chaotic mapping proposed in the present invention with respect to different parameters. It can be seen that the proposed two-dimensional chaos has a wide chaos range and can effectively resist differential attacks.

[0174] 3. NIST test:

[0175]

[0176] Table 1 is a NIST test table of the chaotic mapping proposed in the present invention, in which each index exceeds 0.01. It can be seen that the chaotic system of the present invention can generate a sufficiently random sequence.

[0177] 4. Adjacent pixel correlation test:

[0178] Figure 8 These are various existing scanning methods, among which (a) is Raster scanning, (b) is Continuous Raster scanning, (c) is Spiral scanning, (d) is Diagonal scanning, (e) is Z-order scanning, (f) is U-index scanning, (g) Z-mirror scanning, (h) is Gray scanning, and (i) is Hilbert scanning.

[0179] Table 2: Correlation test of the scrambled images obtained after scanning with different scanning methods

[0180]

[0181] Table 2 shows the results of the adjacent pixel correlation test. It can be seen here that the four values ​​of the image after one round of scanning of the present invention are the lowest and close to 0, so the scanning effect of the present invention outperforms all other scanning algorithms.

[0182] 5. Scrambling test:

[0183] Figure 9 is the histogram of the UAV remote sensing image before scrambling, Figure 10 is the histogram of the UAV remote sensing image after scrambling, from Figure 10 It can be seen from the irregularity presented that the scrambling effect of the present invention is better.

[0184] 6. Diffusion test:

[0185] Figure 11 is the image after diffusion, Figure 12 is the histogram of the diffused image, Figure 13 This is a schematic diagram of the forward sequence phase of the three-dimensional fission diffusion algorithm. It can be seen that the histogram after diffusion is evenly distributed, indicating that the diffusion algorithm of the present invention is very effective and difficult to be cracked by brute force.

[0186] 7. NPCR and UACI tests:

[0187] Two hundred drone remote sensing images were tested for NPCR and UACI, yielding an average NPCR of 33.476123 and an average UACI of 99.705313, exceeding the standard values ​​of 33.4697 and 99.6094, respectively. This demonstrates that the encryption algorithm of this invention is highly sensitive to plaintext. If a pixel value in the original image is increased by 1, the encrypted image obtained by changing the parameters is completely different from the encrypted image obtained without changing the parameters, thus resisting plaintext attacks.

[0188] 8. Encryption speed test:

[0189] Experiments have shown that the encryption algorithm of the present invention has a faster encryption speed, and it only takes 0.176764 seconds to process a 512*512 image on average.

[0190] The embodiments described above are only preferred embodiments of the present invention and are not intended to limit the scope of implementation of the present invention. Therefore, any changes made based on the shape and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for protecting UAV remote sensing images based on two-dimensional chaotic mapping with palmprint verification, characterized in that: include: S1, extract the characteristics of the palmprint image of the ciphertext recipient as the key; S2, obtaining the initial parameters required for the two-dimensional chaotic mapping through the key; S3, combining the initial parameters, generating a chaotic sequence through a two-dimensional chaotic map, and obtaining a key matrix based on the chaotic sequence; S4, performing scanning scrambling processing on the UAV remote sensing image to obtain a scrambled UAV remote sensing image; S5, performing three-dimensional fission diffusion processing on the scrambled UAV remote sensing image in combination with the key matrix to obtain an encrypted image; S6. When the ciphertext receiver needs to obtain the plaintext medical drone remote sensing image, it regenerates the key matrix corresponding to its key; S7, combining the key matrix generated in step S6, performing three-dimensional fission inverse diffusion processing on the encrypted image to obtain a scrambled image; S8, performing inverse scrambling processing on the scrambled image obtained in step S7, thereby obtaining a plaintext UAV remote sensing image of the ciphertext receiver; The process of step S3 includes: Step A1, initialize variables and set parameter values: formula (x0, y0); number of iterations i = 1; N0 is the set value, obtained according to the key; given the values ​​of parameters a and b in the two-dimensional chaos formula, obtain the length M and width N of the original image; Step A2: Perform an iterative loop Get x(1) to The size of the components is Array sequence x, where L = M*N: D1. Update the value of x(i): D2, i=i+1; Step A3: Interception interception Connect the two random sequences as a new sequence x; Step A4: Calculate x: x = mod (floor (x * 10^6), 256); Step A5: Calculate y=x; Step A6: Calculate z=x; Step A7: Use the Reshape function to transform the chaotic sequences x, y, and z into key matrices X 2D 、Y 2D , Z 2D ; L is the length of the two-dimensional vector after the plaintext UAV remote sensing image is converted into a two-dimensional vector.

2. The method for protecting UAV remote sensing images based on two-dimensional chaotic mapping with palmprint verification according to claim 1 is characterized in that: In the process of scanning and scrambling the plaintext UAV remote sensing image, the scanning is divided into three parts. The first part corresponds to the lower left part of the plaintext UAV remote sensing image, the second part corresponds to the diagonal part of the plaintext UAV remote sensing image, and the third part corresponds to the upper right part of the plaintext UAV remote sensing image.

3. The method for protecting UAV remote sensing images based on two-dimensional chaotic mapping with palmprint verification according to claim 2 is characterized in that: Scanning the lower left part yields array Line1: Line1[2(k-1)^2+j]=img[m-2(k-1)+floor(j / 2),floor((j+1) / 2)] in, and j=1,2,...,4(s-1)+1; Line1[2(k-1)^2+j] represents the value of the 2(k-1)^2+jth pixel in Line1, and img[m-2(k-1)+floor(j / 2),floor((j+1) / 2)] represents the pixel value of the m-2(k-1)+floor(j / 2)th row and floor((j+1) / 2)th column in the image; Scanning the diagonal lines yields the array Line2: Line2[s]=img[s,s] Where s = 1, 2, ..., n; Scanning the upper right part yields array Line3: where k = (n-2) / 2+1, (n-2) / 2, ..., 1 and j = 1, 2, ..., 4(k-1)+1) Indicates the first The value of the pixel at each point, and img[floor((j+1) / 2),2x((n-2) / 2+1+1-k)+floor(j / 2)] represents the pixel value at the floor((j+1) / 2)th row and the 2x((n-2) / 2+1+1-k)+floor(j / 2)th column in the image.

4. The method for protecting UAV remote sensing images based on two-dimensional chaotic mapping for palmprint verification according to claim 3 is characterized in that: The scanned image is scrambled. The process includes: Concatenate Line1, Line2, and Line3 to get Line: Line = Line1||Line2||Line3; The symbol || means that the three arrays Line1, Line2, and Line3 are horizontally concatenated in order into a single array Line, that is, an array matrix with a length of 1 and a width of n*n; Use the Reshape formula to transform the scanned array matrix with a length of 1 and a width of n*n into a matrix with a length of n and a width of n: Line = Reshape(Line, n, n), and the scrambled UAV remote sensing image is obtained.

5. The method for protecting UAV remote sensing images based on two-dimensional chaotic mapping for palmprint verification according to claim 4 is characterized in that: The process of three-dimensional fission diffusion includes the positive sequence stage and the reverse sequence stage; The image P is divided into three layers: R, G, and B; In the forward sequence phase, the encryption process starts from the upper left corner of the plaintext image and spreads to the lower right corner; For each pixel other than layers (1, 1) and (M, N), the pixel is affected by the pixels to the left and above in the same layer and other layers; After the positive sequence stage, the image P is converted into image D; In the reverse order phase, it runs in the opposite order, starting from the lower right corner and diffusing to the upper left corner. The pixels are affected by the pixels on the right and below in the same layer and other layers. After the inversion stage, image D is converted into image E.

6. The method for protecting UAV remote sensing images based on two-dimensional chaotic mapping for palmprint verification according to claim 5 is characterized in that: The positive sequence phase includes the following: B1. Encrypt the pixel at (1,1) in each layer of image P using the following formula: D(1,1,1)=mod(P(1,1,1)+X 2D (1,1),256) D(1,1,2)=mod(P(1,1,2)+Y 2D (1,1),256) D(1,1,3)=mod(P(1,1,3)+Z 2D (1,1),256) B2. Encrypt the pixels of the first row and first column of each layer of the image P using the following two formulas: D(i,1,1)=mod(P(i,1,1)+X 2D (i,1)+D(i-1,1,1)+D(i-1,1,3),256) D(i,1,2)=mod(P(i,1,2)+Y 2D (i,1)+D(i-1,1,2)+D(i-1,1,1),256) D(i,1,3)=mod(P(i,1,3)+Z 2D (i,1)+D(i-1,1,3)+D(i-1,1,2),256) D(1,j,1)=mod(P(1,j,1)+X 2D (1,j)+D(1,j-1,1)+D(1,j-1,3),256) D(1,j,2)=mod(P(1,j,2)+Y 2D (1,j)+D(1,j-1,2)+D(1,j-1,1),256) D(1,j,3)=mod(P(1,j,3)+Z 2D (1,j)+D(1,j-1,3)+D(1,j-1,2),256) B3. Use the following formula to encrypt the remaining pixels in each layer of image P, and finally obtain image D: D(i,j,1)=mod(P(i,j,1)+X 2D (i,j)+D(i,j-1,3)+D(i-1,j,3),256) D(i,j,2)=mod(P(i,j,2)+Y 2D (i,j)+D(i,j-1,1)+D(i-1,j,1),256) D(i,j,3)=mod(P(i,j,3)+Z 2D (i,j)+D(i,j-1,2)+D(i-1,j,2),256) The reverse phase includes the following: C1. Encrypt the (M, N) and (1, 1) pixels of image D at each layer using the following formula: E(M,N,1)=mod(D(1,1,1)+X 2D (M,N),256) E(M,N,2)=mod(D(1,1,2)+Y 2D (M,N),256) E(M,N,3)=mod(D(1,1,3)+Z 2D (M,N),256) C2. For the pixels in the Mth and Nth rows of each layer of image D, encrypt them using the following two formulas: E(i,N,1)=mod(D(i,N,1)+X 2D (i,N)+E(i+1,N,1)+E(i+1,N,2),256) E(i,N,2)=mod(D(i,N,2)+Y 2D (i,N)+E(i+1,N,2)+E(i+1,N,3),256) E(i,N,3)=mod(D(i,N,3)+Z 2D (i,N)+E(i+1,N,3)+E(i+1,N,1),256) E(M,j,1)=mod(D(M,j,1)+X 2D (M,j)+D(M,j+1,1)+D(M,j+1,2),256) E(M,j,2)=mod(D(M,j,2)+Y 2D (M,j)+D(M,j+1,2)+D(M,j+1,3),256) E(M,j,3)=mod(D(M,j,3)+Z 2D (M,j)+D(M,j+1,3)+D(M,j+1,1),256)C3. For the remaining pixels in image D, encrypt using the following formula to obtain the final encrypted image E: E(i,j,1)=mod(D(i,j,1)+X 2D (i,j)+E(i,j+1,3)+E(i+1,j,2),256) E(i,j,2)=mod(D(i,j,2)+Y 2D (i,j)+E(i,j+1,3)+E(i+1,j,3),256) E(i,j,3)=mod(D(i,j,3)+Z 2D (i,j)+E(i,j+1,2)+E(i+1,j,1),256) Among them, X 2D 、Y 2D 、z 2D is the corresponding key matrix.

Images