A crosstalk-free noise-free multi-image encryption method based on computational ghost imaging

Abstract

This invention proposes a crosstalk-free multi-image encryption method based on computational ghost imaging. The method first encrypts multiple plaintext images using a computational ghost imaging system and binarizes the bucket detector value sequences to obtain multiple sets of binary data sequences of the same length. Then, the obtained sets of binary data sequences are converted into a grayscale image with decimal pixel values. Finally, a pixel bit position scrambling algorithm is used to scramble the pixel values ​​of the grayscale image to obtain the final ciphertext image. During decryption, the multiple sets of binary data sequences corresponding to the multiple plaintext images are recovered through the reverse pixel bit position scrambling operation. Then, legitimate users use the key to decrypt their respective plaintext images, and the decrypted image is unaffected by crosstalk noise from other images. This invention solves the crosstalk noise problem in optical multi-image encryption and increases the encryption capacity, which is beneficial for improving the transmission and storage efficiency of multi-image information.

Description

Technical Field

[0001] The present invention relates to the field of information security technology, and in particular to a crosstalk-free noise-free multi-image encryption method based on computational ghost imaging. Background Art

[0002] In today's era, the rapid development of information technology has provided great convenience for people's lives and work, but at the same time, information security issues have become very critical. As an emerging encryption technology, optical encryption technology has attracted more and more attention from researchers for its high-speed parallel processing and multi-dimensional simultaneous processing capabilities of information. In 1995, Refragier and Javidi first used the optical 4f system to realize double random phase encoding. Subsequently, many different optical encryption methods such as Fresnel transform, fractional Fourier transform, diffraction imaging, joint transform correlator, and computational ghost imaging have been proposed.

[0003] At present, with the rapid development of big data technology and the continuous enhancement of information transmission capabilities, single image encryption technology can no longer meet people's daily needs for information. In order to improve the efficiency of information transmission, multi-image encryption technology has attracted more and more attention from scholars.

[0004] There are two main methods for multi-image encryption. One is to combine multiple plaintext images into one image through compression and other methods, and then send it to the optical encryption system for encryption. During the decryption process, the original multiple images are restored through the inverse process of encryption such as spatial multiplexing technology. This method usually loses some image information because it requires compression or sparse transformation of the plaintext image. In addition, due to the correlation between plaintext images, all plaintext images will be decrypted at the same time, and it is impossible for each user to extract only the required part of the information. Another method is to use an encryption system to encrypt each plaintext image separately, and then superimpose them to obtain ciphertext. When decrypting, different keys can be used to decrypt any plaintext image that needs to be extracted separately through wavelength multiplexing technology, position multiplexing technology, and angle multiplexing technology. Obviously, this method is more in line with the needs of practical applications. However, since the ciphertext is directly superimposed, when decrypting one of the images, it will inevitably be affected by the crosstalk noise of other images, resulting in a very low encryption capacity. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a crosstalk noise-free multi-image encryption method based on computational ghost imaging. The present invention can solve the crosstalk noise problem of the existing multi-image encryption method and greatly improve the encryption capacity in the multi-image encryption process.

[0006] The specific steps of the present invention are as follows:

[0007] Encryption process:

[0008] Step 1: Use the computational ghost imaging system to perform the same number of initial encryptions on multiple plaintext images, and binarize the encrypted bucket detector values ​​to obtain multiple sets of binary data sequences of the same length.

[0009] Step 2: Convert the multiple sets of binary data sequences obtained in step 1 into a grayscale image with pixel values ​​as decimal numbers.

[0010] Step 3: Use the designed pixel bit position scrambling algorithm to scramble the pixel bits in the grayscale image obtained in step 2 to obtain the final ciphertext image, and the encryption is completed.

[0011] Decryption process:

[0012] According to the designed pixel bit position scrambling algorithm, the reverse operation is performed to restore the grayscale image in step 2 from the final ciphertext image; the grayscale value of each pixel of the grayscale image is represented by a binary number, and the reverse operation is performed according to the process of step 2 to obtain multiple groups of binary data sequences of the same length. These binary data sequences are the calculated ghost imaging bucket detector values ​​after binarization; then the legitimate user can use his own key to decrypt a group of binary data sequences according to the reverse operation of the calculated ghost imaging encryption to restore the corresponding image; because each group of binary data sequences is independent of each other, the restored image is not affected by the crosstalk noise of other images.

[0013] Preferably, the binarization threshold C in the encryption step 1 is determined by the following formula:

[0014]

[0015] Where B is the bucket detector value corresponding to each set of data, and max and min represent the maximum and minimum value operations respectively.

[0016] Preferably, the method for converting multiple sets of binary data sequences into a grayscale image with pixel values ​​as decimal numbers in the encryption step 2 is: take one bit (0 or 1) from each of the multiple sets of binary data sequences, and combine them into a binary number in the order of group numbers, and the number of bits of this binary number is the number of plaintext images processed; calculate the decimal number corresponding to this binary number, and record this decimal number as a pixel grayscale value; then take one bit from each of the multiple sets of binary data sequences again, construct another pixel grayscale value and record it; repeat this cycle until the multiple sets of binary data sequences are taken; these constructed grayscale values ​​constitute a grayscale image with pixel values ​​as decimal numbers. .

[0017] Preferably, the pixel bit position scrambling algorithm designed in the encryption step three refers to using different random sequences to scramble each bit of the corresponding binary number for each pixel gray value of the image in turn. For example, assuming that the gray value of a certain pixel is 179, the corresponding binary number is 10110011. If the bits of the binary number are arranged and scrambled according to a given random sequence 62857431, that is, the original "0" of the 6th bit is placed at the 1st bit, the original "0" of the 2nd bit is placed at the 2nd bit..., the original "1" of the 7th bit is placed at the 5th bit..., the result is 00101111, and the corresponding decimal number is 47, which is completely different from the original pixel value 179.

[0018] Preferably, the random sequence required by the pixel bit position randomization algorithm in the encryption step three comes from a preset key library. For example, assuming that in the encryption step two, the binary data is to be combined into a grayscale image with a grayscale value between 0 and 255, since each pixel value is represented by an 8-bit binary number, the binary data sequence corresponding to the 8 plaintext images can be represented as an array [1,2,3,4,5,6,7,8], and the sequence of their serial numbers can be combined at will to obtain 40320 groups of random sequences; then these random sequences are numbered from 1 to 40320, so that when extracting the random sequence, the corresponding number can be accessed, that is, the number is the key used, without the need to transmit a large number of random sequences, which not only greatly reduces the amount of key transmission, but also has a certain degree of confusion.

[0019] Compared with the existing methods, the present invention has the following beneficial effects:

[0020] (1) The present invention realizes a multi-image encryption method without crosstalk noise, that is, the decrypted image will not be affected by the crosstalk noise of other images;

[0021] (2) Since the present invention combines and scrambles the data after the initial encrypted ciphertext is binarized into the final ciphertext, the encryption capacity of the present invention will be greatly improved as the bit depth of the decimal number varies, and the quality of the encrypted image will not change with the increase in the number of images.

[0022] (3) If the ciphertext of the present invention is combined into higher-base data or color ciphertext, the encryption capacity will continue to increase. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 (a)-(d) are binary images; (e)-(h) are grayscale images.

[0024] Figure 2 Figure 2 shows the multi-image encryption system in the present invention. (a) Encryption process; (b) Decryption process.

[0025] Figure 3 (a) is the principle diagram of the pixel layer scrambling algorithm; (b) is a diagram of the pixel layer scrambling algorithm applied to the image scrambling process.

[0026] Figure 4 is the encrypted ciphertext image.

[0027] Figure 5 (a)-(h) are the images decrypted by the proposed encryption system.

[0028] Figure 6 These are the robust attack results: (a) Gaussian noise attack result; (b) salt and pepper noise attack result; (c) speckle noise attack result; (d) cropping attack result. DETAILED DESCRIPTION

[0029] In order to make the objectives, technical solutions and advantages of the present invention more clear, a specific embodiment of the present invention is described below in conjunction with the accompanying drawings and specific implementation methods.

[0030] In this example, four binary images and four grayscale images of 64×64 pixels are used as the images to be encrypted. Figure 1 .

[0031] Encryption process: 8 plaintext images are encrypted using Figure 2 (a) is used for encryption, and the number of measurements is 10,000, and 8 groups of bucket detector values ​​of the same length are obtained {B i}, i is the sequence number of the image, ranging from 1 to 8.

[0032] For 8 groups of bucket detector values ​​{B i} Perform binarization processing, the binarization threshold C i are the average values ​​of the maximum and minimum values ​​of each group of bucket detector values, respectively, as shown in formula (1), where max and min are the maximum and minimum value operations. The result of binarization is 8 groups of binary data {D i}.

[0033]

[0034] From 8 sets of binary data sequences {D i}, and form an 8-bit binary number according to the sequence of each group number. The corresponding decimal number must be in the range of 0-255. This decimal number is used as a pixel grayscale value and recorded; then, the 8 groups of binary data sequences {D i}, and construct a pixel grayscale value and record it; this cycle repeats until 8 sets of binary data sequences {D i}After taking; these constructed grayscale values ​​constitute a grayscale image whose pixel values ​​are decimal numbers.

[0035] To enhance security, we designed a pixel bit position scrambling algorithm to scramble grayscale images. The specific process is as follows: Figure 3 As shown in (a), for each pixel gray value, an 8-bit random sequence is used to scramble each bit of the corresponding binary number. Assuming that the binary number corresponding to a certain pixel gray value is 10110011, the random sequence 62857431 is used to scramble the bit positions of this binary number, and the result is 00101111.

[0036] In order to reduce the amount of key transmission, the random sequence of the designed pixel bit layer scrambling algorithm is selected from the pre-agreed key library. For the decimal number corresponding to the 8-bit binary number, the key library is a total of 40320 groups of random sequences obtained by random arrangement and combination of the array [1,2,3,4,5,6,7,8], which are numbered from 1 to 40320. At this time, the random sequence can be obtained by accessing the number corresponding to the random sequence, so that when transmitting, both parties only need to transmit the number corresponding to the random sequence.

[0037] like Figure 3 As shown in (b), the specific process of applying the pixel bit position scrambling algorithm is that, for a 3×3 pixel image, each pixel grayscale value is first represented by an 8-bit binary number, and then the pixel bit position scrambling algorithm is used to randomly scramble the order of each bit of the 8-bit binary number. Finally, the 8-bit binary number with the scrambled bit order is converted back into the corresponding decimal number to obtain Figure 4 The final ciphertext image is shown.

[0038] Decryption process: Figure 2 As shown in (b), the grayscale value of each pixel of the ciphertext image is operated according to the reverse pixel bit position scrambling algorithm, and then the binary numbers corresponding to all pixel values ​​are recombined according to each bit position to form 8 groups of binary data sequences {D i}, the legitimate user can use his own key to decrypt a set of binary data sequences according to the reverse operation of computing ghost imaging encryption and restore the corresponding image, such as Figure 5 As shown, since the eight sets of binary data sequences are independent of each other, the restored image is not affected by the crosstalk noise of the other seven images.

[0039] In the present invention, the correlation coefficient (CC) and the peak signal-to-noise ratio (PSNR) are used as the criteria for evaluating the quality of the decrypted image and the original plaintext image.

[0040]

[0041] Formula (2) is the calculation formula of CC value, where A represents the decrypted image and B represents the original plaintext image. m and n are the length and width of the image respectively. Represents the average operation of X.

[0042]

[0043] Formula (3) is the calculation formula of PSNR, where g′(x, y) represents the decrypted image and g(x, y) represents the original plaintext image. X×Y and k are the size and grayscale of the image respectively.

[0044] The CC value and PSNR of the decrypted image are shown in Table 1:

[0045] Table 1 CC value and PSNR of decrypted images

[0046]

[0047] It can be seen that the decrypted images can be distinguished regardless of whether they are binary images or grayscale images due to the absence of crosstalk noise. The CC values ​​of all decrypted images are above 0.7, the PSNR of binary images are around 10dB, and the PSNR of grayscale images are above 14dB.

[0048] Robustness analysis: Figure 6 For the convenience of explaining the problem, we use Figure 1 The binary image in (b) and Figure 1 Take the grayscale image in (f) as an example, Figure 4 The ciphertext image is added with Gaussian noise with mean 0 and variance 0.001, salt and pepper noise with density 0.5, speckle noise with mean 0 and variance 0.001, and 50% of the part is cropped, and then decrypted. The results show that the CC of the decrypted image is above 0.36, which proves that the present invention has strong resistance to robust attacks, that is, during the transmission of the ciphertext image, if it is attacked by noise or part of the information is contaminated, the original encrypted image can still be decrypted.

[0049] Encryption capacity analysis: In this example, we take the decimal number corresponding to an 8-bit binary number as an example. If the ciphertext is a grayscale image, 8 images can be encrypted. It is easy to imagine that if the ciphertext is a color image, 24 images can be encrypted. If the pixel depth of the decimal number is increased or other bases are used to represent it, more images can be encrypted.

[0050] Security analysis: As we all know, the traditional encryption system based on computational ghost imaging is difficult to resist known plaintext attack (KPA), chosen plaintext attack (CPA), and even ciphertext-only attack (COA) due to the linear relationship and role symmetry between plaintext and ciphertext. However, the method proposed in this paper breaks this balance, because the ciphertext has been binarized and operated by the pixel-level scrambling algorithm, so that the relationship between plaintext and ciphertext is no longer a simple correspondence. Therefore, for the method proposed in this paper, the existing KPA and CPA are invalid. In COA, the attacker takes advantage of the same statistical characteristics of the ciphertext corresponding to the same type of plaintext image. Here, the pixel-level scrambling algorithm makes the plaintext and ciphertext no longer correspond one-to-one, so the modulation matrix cannot be obtained through the statistical characteristics of the ciphertext.

[0051] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A crosstalk-free noise-free multi-image encryption method based on computational ghost imaging, characterized in that: The encryption steps include: S1: After multiple plaintext images are initially encrypted with the same number of measurements using a computational ghost imaging system, their respective bucket detector values ​​are obtained, and then binarization is performed to obtain multiple sets of binary data sequences of the same length; S2: Take the number of binary data sequences obtained in S1 as the bit depth, take one bit from each sequence in turn to form a binary number, and convert the binary number into a decimal number, record it as the pixel grayscale value, repeat this process until all the bits in each sequence are taken, and then construct a grayscale image with pixel values ​​in decimal. S3: The grayscale image in S2 is subjected to the designed pixel bit position scrambling algorithm, and the bit position of each pixel is scrambled according to different random sequences, so that the pixels of the image present a state of random and chaotic values, and the final ciphertext image is obtained, completing the encryption process; And the decryption steps: T1: Perform the reverse operation according to the designed pixel bit position scrambling algorithm to restore the grayscale image in S2 from the final ciphertext image; T2: Represent the grayscale value of each pixel of the grayscale image obtained in T1 with a binary number, and perform the reverse operation according to S2 to obtain multiple sets of binary data sequences of the same length; T3: The legitimate user uses his own key to decrypt a set of binary data sequences according to the reverse operation of calculating ghost imaging encryption and restore the corresponding image.

2. According to claim 1, a method for crosstalk-free noise-free multi-image encryption based on computational ghost imaging is characterized in that: In step S1, the obtained bucket detector value is binarized, that is, the average value C of the maximum value and the minimum value in all bucket detector values ​​B is first calculated: Among them, max and min represent the maximum and minimum operations respectively, and then C is used as the threshold for binarization processing. The data greater than C is set to 1, and the part less than C is set to 0, and the binarized data sequence D is obtained.

3. According to the method of claim 1, the method is characterized in that: The random sequence required by the pixel bit position scrambling algorithm in step S3 comes from the key library. According to the determined decimal bit depth, a random sequence of the corresponding number of groups is obtained, and then these random sequences are numbered. When extracting the random sequence, the random sequence is obtained by accessing the number, that is, the number is used as the key, which not only greatly reduces the transmission amount of the key, but also has a certain degree of confusion.

Images