Face image transmission method and system for security door

Abstract

The present application relates to the field of image transmission, and more particularly to a face image transmission method and system for security door. The method comprises the steps of: segmenting a face image into a plurality of image blocks, and performing a plurality of rounds of exchange on the image blocks until the degree of damage to the topological relationship of the face image after exchange is greater than a preset damage degree threshold, thereby obtaining a final exchanged face image; obtaining a key sequence of each gray co-occurrence pair according to the statistical values of each gray co-occurrence pair in the gray co-occurrence matrix of the final exchanged face image, adding the offset values of each element in the key sequence to the mean values of each gray co-occurrence pair in turn, adding the span adjustment values of each element in the key sequence to the span values of each gray co-occurrence pair in turn to obtain adjusted gray co-occurrence pairs, and adjusting the pixel values of each pixel in the final exchanged face image based on the adjusted gray co-occurrence pairs to obtain an encrypted face image, thereby achieving transmission processing. The face encryption effect is improved, and the transmission security is ensured.

Description

Technical Field

[0001] This invention relates to the field of image transmission, and in particular to a method and system for transmitting facial images for security gates. Background Technology

[0002] With the widespread application of facial recognition technology in public safety, security gates, as crucial security equipment in densely populated areas, are increasingly becoming a core means of identity verification and security screening due to their facial image acquisition and transmission functions. However, facial images, as important carriers of biometric data, contain sensitive personal information. If leaked during transmission, they pose risks such as identity theft and privacy violations.

[0003] Currently, security measures for facial image transmission at security gates have significant limitations. On the one hand, traditional encryption algorithms often employ a holistic encryption strategy, which, while ensuring data confidentiality, suffers from high computational complexity, making it difficult to meet the high-concurrency, low-latency real-time transmission requirements of security gates. On the other hand, some lightweight encryption schemes only perform simple transformations on image pixel values, failing to effectively defend against feature extraction-based attacks. The topological relationships of facial images (i.e., macroscopic geometric features such as facial feature positions and contour structures) and detailed texture information (such as microscopic features like skin texture and wrinkle distribution) are crucial for facial recognition, yet existing technologies often neglect the differentiated protection of these two types of features. For example, single pixel-level encryption cannot alter the image's topological structure, allowing attackers to reconstruct facial features through geometric analysis; simply adding noise to the texture is also insufficient to resist deep learning-based image restoration techniques, significantly reducing the encryption effectiveness. Therefore, a multi-layered encryption method that considers both topological structure and detailed texture is urgently needed to construct a security protection system from both macroscopic and microscopic dimensions, thereby improving the security of facial images during transmission at security gates. Summary of the Invention

[0004] To address the challenge of improving facial image security through encryption at both macro and micro levels, this invention provides a facial image transmission method and system for security gates.

[0005] In a first aspect, the present invention provides a method for transmitting facial images for security gates, employing the following technical solution:

[0006] A method for transmitting facial images for use in security gates, comprising the following steps:

[0007] Acquire facial images;

[0008] A face image is divided into several image blocks, and the image blocks are swapped several times until the degree of topological relationship destruction of the swapped face image is greater than a preset destruction threshold, thus obtaining the final swapped face image. The degree of topological relationship destruction reflects the difference between the positional relationship of feature points in the swapped face image and the positional relationship of human feature points in the original face image.

[0009] The gray-level co-occurrence matrix of the final swapped face image is obtained. Based on the statistical values ​​of each gray-level co-occurrence pair in the gray-level co-occurrence matrix, a key sequence for each gray-level co-occurrence pair is obtained. Each element in the key sequence contains an offset value and a span adjustment value. The mean of each gray-level co-occurrence pair is sequentially added to the offset value of each element in the key sequence, and the span value of each gray-level co-occurrence pair is sequentially added to the span adjustment value of each element in the key sequence to obtain the adjusted gray-level co-occurrence pair. Based on the adjusted gray-level co-occurrence pair, the pixel values ​​of each pixel in the final swapped face image are adjusted to obtain the encrypted face image, thus achieving transmission processing.

[0010] This invention alters the relative positions of face images by swapping image blocks, effectively disrupting the topological relationships of the face and thus macroscopically destroying facial information. Furthermore, by adjusting the span values ​​of gray-level co-occurrence pairs in the swapped face images, the original gradient information is altered, thereby better masking the texture information in the face images. Even further, by adjusting the mean value of the gray-level co-occurrence pairs in the swapped face images, the gray values ​​in the face images deviate from the original gray values, thus better masking the gray value information in the face images.

[0011] Preferably, the step of segmenting the face image into several image blocks includes:

[0012] The face image is evenly divided into a preset first number of image blocks.

[0013] Preferably, the step of performing several rounds of image block swapping until the degree of topological disruption of the swapped face image exceeds a preset disruption threshold, to obtain the final swapped face image, includes:

[0014] First chaotic data is generated. Based on the first chaotic data, a preset second number of image blocks and a preset second number of image blocks are extracted from all image blocks as active exchange image blocks and passive exchange image blocks, respectively. The active exchange image blocks and passive exchange image blocks are exchanged to obtain a face image after one round of exchange. The degree of topological relationship destruction is calculated based on the face image after one round of exchange. In response to the degree of topological relationship destruction not being greater than a preset destruction degree threshold, a second round of exchange is performed on each image block in the face image after one round of exchange. In response to the degree of topological relationship destruction being greater than the preset destruction degree threshold, the obtained exchanged face image is recorded as the final exchanged face image.

[0015] This invention uses the degree of topological relationship destruction as the cutoff condition for exchanging image blocks, thereby preventing insufficient number of exchanges from causing the topological relationship of the face to be not destroyed well, or preventing excessive number of exchanges from causing useless workload.

[0016] Preferably, the step of calculating the degree of topological relationship disruption based on the face images after one round of swapping includes:

[0017] Obtain the corresponding feature points of each feature point in the face image after one round of swapping;

[0018] In the face image, every two feature points are formed into a feature point pair. The corresponding feature point pair in the face image after one round of swapping is obtained. The vector formed by the two feature points in each feature point pair is recorded as the description vector. The degree of destruction of each feature point pair is obtained by multiplying the absolute value of the difference between the magnitude of each feature point pair and the description vector of the corresponding feature point pair by the included angle value.

[0019] The normalized value of the mean of the degree of destruction of all feature point pairs is denoted as the degree of topological relationship destruction.

[0020] This invention accurately determines the changes in the topological relationships of a face image caused by image block swapping by analyzing the relative positional changes of feature points in the swapped face image compared to the corresponding feature points in the face image before the swap.

[0021] Preferably, obtaining the gray-level co-occurrence matrix of the final swapped face image includes:

[0022] In any direction, the gray values ​​of every two adjacent pixels in each channel of the final swapped face image are used to form a gray-level co-occurrence pair. The gray-level co-occurrence matrix is ​​obtained by statistically analyzing the gray-level co-occurrence pairs obtained in each channel image.

[0023] Preferably, obtaining the key sequence of each gray-level co-occurrence pair based on the statistical value of each gray-level co-occurrence pair in the gray-level co-occurrence matrix includes:

[0024] The number of elements in the key sequence of each gray-level co-occurrence pair is calculated based on the occurrence frequency of each gray-level co-occurrence pair.

[0025] Generate some chaotic data and denote it as the second chaotic data. Take the vector formed by two second chaotic data as an element. Denote the sequence formed by L elements as the key sequence of each gray-level co-occurrence pair. L represents the number of elements in the key sequence of each gray-level co-occurrence pair. Denote the first second chaotic data of each element in the key sequence as the offset. Denote the second second chaotic data of each element in the key sequence as the span adjustment value.

[0026] Preferably, the step of calculating the number of elements in the key sequence of each gray-level co-occurrence pair based on the occurrence frequency of each gray-level co-occurrence pair includes:

[0027] The number of elements in the key sequence of each gray-level co-occurrence pair is obtained by multiplying the normalized value of the occurrence frequency of the gray-level co-occurrence pair by a preset adjustment value and then rounding it up.

[0028] This invention assigns different numbers of keys to gray-level co-occurrence pairs with different frequencies of occurrence based on the frequency of occurrence of the gray-level co-occurrence pairs. This results in more keys being assigned to gray-level co-occurrence pairs with high frequency of occurrence, increasing the complexity of the keys and thus better masking the original information of the gray-level co-occurrence pairs with high frequency of occurrence, thereby increasing the difficulty of cracking.

[0029] Preferably, the method for obtaining the adjusted gray-level co-occurrence pairs includes:

[0030] The adjusted center of the gray-level co-occurrence pair is obtained by adding the mean of the two gray values ​​in the gray-level co-occurrence pair to the offset value of the first element in the key sequence. The adjusted span value is obtained by adding the span adjustment value of the first element to the span value of the gray-level co-occurrence pair. The adjusted first data of the gray-level co-occurrence pair is obtained by subtracting half of the adjusted span value from the adjusted center. The adjusted second data of the gray-level co-occurrence pair is obtained by adding half of the adjusted span value to the adjusted center. The data pair formed by the adjusted first data and the adjusted second data is used as the gray-level co-occurrence pair after the first round of encryption. The adjusted gray-level co-occurrence pair is obtained by performing L rounds of encryption on the gray-level co-occurrence pair using each element in the key sequence.

[0031] This invention adjusts the span value and the mean value of gray-level co-occurrence pairs based on the key sequence, thereby effectively changing the original value of gray-level and gradient information, thus effectively masking the texture information in face images.

[0032] Preferably, the step of adjusting the pixel values ​​of each pixel in the finally swapped face image based on the adjusted grayscale co-occurrence pair to obtain the encrypted face image includes:

[0033] The encrypted face image is obtained by replacing the gray values ​​of the corresponding pixels in each channel of the final swapped face image with the gray values ​​in the adjusted gray-level co-occurrence pairs.

[0034] Secondly, the present invention provides a facial image transmission system for security gates, employing the following technical solution:

[0035] A facial image transmission system for security gates includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the aforementioned facial image transmission method for security gates is implemented.

[0036] By adopting the above technical solution, a computer program is generated from the above-mentioned method for transmitting facial images for security gates and stored in a memory so that it can be loaded and executed by a processor. This allows for the creation of a terminal device based on the memory and processor, making it convenient to use.

[0037] The present invention has the following technical effects:

[0038] This invention alters the relative positional relationships in a face image by swapping image blocks, thereby effectively disrupting the topological relationships of the face and thus macroscopically destroying the face information.

[0039] Furthermore, by adjusting the span value of gray-level co-occurrence pairs in the swapped face image, the original gradient information is changed, thereby better masking the texture information in the face image;

[0040] Furthermore, by adjusting the mean information of the gray-level co-occurrence pairs in the swapped face image, the gray-level values ​​in the face image are made to deviate from the original gray-level values, thereby better masking the gray-level information in the face image. Attached Figure Description

[0041] Figure 1 This is a flowchart of a method for transmitting facial images for a security gate according to an embodiment of the present invention. Detailed Implementation

[0042] This invention discloses a method for transmitting facial images for security gates, referring to... Figure 1 This includes steps S1-S4:

[0043] S1: Obtain the face image.

[0044] Specifically, it involves acquiring facial images of people used for security gates.

[0045] S2: Divide the face image into several image blocks, and perform several rounds of swapping on the image blocks until the degree of topological relationship destruction of the swapped face image is greater than a preset destruction threshold, and obtain the final swapped face image. The degree of topological relationship destruction reflects the difference between the positional relationship of feature points in the swapped face image and the positional relationship of feature points in the original face image.

[0046] It should be noted that the topological relationships of a face are crucial information, therefore, better protection of facial information requires disrupting these topological relationships. Topological relationships are primarily manifested in the relative positions of facial feature points; thus, facial information protection can be achieved by altering the relative positions of these feature points.

[0047] S20: Divide the face image into several image blocks.

[0048] Preferably, as an example, the face image is segmented into several image patches, including:

[0049] The face image is evenly divided into a preset first number of image blocks.

[0050] S21: Perform several rounds of swapping on the image blocks until the degree of topological disruption of the swapped face image exceeds a preset disruption threshold, thus obtaining the final swapped face image.

[0051] Preferably, as an example, the image blocks are swapped several times until the degree of topological disruption of the swapped face image exceeds a preset disruption threshold, resulting in the final swapped face image, including:

[0052] Generate some chaotic data and denote it as the first chaotic data. Based on the first chaotic data, extract a preset second number of image blocks and a preset second number of image blocks from all image blocks as active exchange image blocks and passive exchange image blocks, respectively. Exchange the active exchange image blocks and passive exchange image blocks to obtain the face image after one round of exchange.

[0053] Calculate the degree of topological disruption based on the face images after one round of swapping;

[0054] If the degree of topological relationship disruption is not greater than a preset disruption threshold, another round of swapping is performed on each image block in the face image after one round of swapping. If the degree of topological relationship disruption is greater than the preset disruption threshold, the resulting swapped face image is recorded as the final swapped face image.

[0055] To make it easier to understand, the following example illustrates the specific process of exchanging image patches in one round.

[0056] Assuming the second quantity is 5, the generated first chaotic data are 61, 5, 22, 6, 80, 5, 33, 70, 2, and 66 respectively; based on the first 5 first chaotic data, the 61st, 5th, 22nd, 6th, and 80th image blocks are selected as active exchange image blocks, and based on the last 5 first chaotic data, the 5th, 33rd, 70th, 2nd, and 66th image blocks are selected as passive exchange image blocks.

[0057] The first actively swapped image block (the 61st image block) is swapped with the first passively swapped image block (the 5th image block). The second actively swapped image block (the 5th image block) is swapped with the second passively swapped image block (the 33rd image block). The third actively swapped image block (the 22nd image block) is swapped with the third passively swapped image block (the 70th image block). The fourth actively swapped image block (the 6th image block) is swapped with the fourth passively swapped image block (the 2nd image block). The fifth actively swapped image block (the 80th image block) is swapped with the fifth passively swapped image block (the 66th image block). This completes the first round of swaps and yields the face image after one round of swaps.

[0058] It is understandable that each image patch contains some feature points. By swapping the positions of image patches, the relative positions of the feature points can be disrupted, thereby disrupting the topological relationship of the face.

[0059] The above embodiments involve the first chaotic data and the degree of topological relationship destruction. The method for obtaining the first chaotic data and the degree of topological relationship destruction is described below.

[0060] First, we will introduce the method for obtaining the first chaotic data.

[0061] Preferably, as an example, the method for obtaining the first chaotic data includes:

[0062] Obtain the number of image patches, generate chaotic data using Logistic mapping, perform modulo operation on the chaotic data and the number of image patches, and record the resulting data as the first chaotic data.

[0063] Then, the method for obtaining the degree of topological relationship disruption will be introduced.

[0064] Preferably, as an example, methods for obtaining the degree of topological relationship disruption include:

[0065] Obtain the corresponding feature points of each feature point in the face image after one round of swapping;

[0066] In the face image, every two feature points are formed into a feature point pair. The corresponding feature point pair in the face image after one round of swapping is obtained. The vector formed by the two feature points in each feature point pair is recorded as the description vector. The degree of destruction of each feature point pair is obtained by adding the angle value to the absolute value of the difference between the magnitude of each feature point pair and the description vector of the corresponding feature point pair.

[0067] The normalized value of the mean of the degree of destruction of all feature point pairs is denoted as the degree of topological relationship destruction.

[0068] Preferably, as an example, methods for obtaining the degree of topological relationship disruption include:

[0069] The only difference from the alternative embodiment for obtaining the degree of topological relationship disruption is that the degree of disruption of each feature point pair is obtained by multiplying the absolute value of the difference between the magnitude of each feature point pair and the magnitude of the description vector of the corresponding feature point pair by the included angle value.

[0070] It is understandable that by analyzing the difference between the relative positional relationships of the feature points after the swap and the relative positional relationships of the feature points before the swap, the degree of disruption of the topological relationship can be reflected, thereby judging the disruption of the original topological relationship caused by image patch swapping, and then judging whether the desired disorder has been achieved, and whether the swapping process should continue.

[0071] S3: Obtain the gray-level co-occurrence matrix of the final swapped face image. Obtain the key sequence of each gray-level co-occurrence pair based on the statistical value of each gray-level co-occurrence pair in the gray-level co-occurrence matrix. Each element in the key sequence contains an offset value and a span adjustment value. Add the offset value of each element in the key sequence to the mean of each gray-level co-occurrence pair in turn. Add the span adjustment value of each element in the key sequence to the span value of each gray-level co-occurrence pair in turn to obtain the adjusted gray-level co-occurrence pair. Adjust the pixel value of each pixel in the final swapped face image based on the adjusted gray-level co-occurrence pair to obtain the encrypted face image, so as to realize the transmission processing.

[0072] It should be noted that while image patch swapping in step S2 can effectively disrupt the topological relationships of the face, this primarily disrupts the macroscopic information of the face. Detailed texture information is also crucial for the face; therefore, to better protect facial information, it is necessary to disrupt this detailed texture information as well. The gray-level co-occurrence matrix (GLCM) can effectively reflect the detailed texture information of the face; thus, by disrupting the gray-level co-occurrence pair information of the GLCM, the detailed texture information of the face can be effectively disrupted.

[0073] S30: Obtain the gray-level co-occurrence matrix of the final swapped face image.

[0074] Preferably, as an example, obtaining the gray-level co-occurrence matrix of the final swapped face image includes:

[0075] Obtain each channel image of the final swapped face image.

[0076] In any direction, the gray values ​​of every two adjacent pixels in each channel image are used to form a gray-level co-occurrence pair. The gray-level co-occurrence matrix is ​​obtained by statistically analyzing the gray-level co-occurrence pairs obtained in each channel image. Any direction can be horizontal, vertical, or at a 45-degree angle to the horizontal direction; its specific direction is not restricted.

[0077] Assume the generated gray-level co-occurrence matrix is ​​as shown in Table 1: the data in the first column represents the first gray value in each gray-level co-occurrence pair, the data in the first row represents the second gray value in each gray-level co-occurrence pair, and the data in the remaining cells represent the frequency of occurrence of each gray-level co-occurrence pair. For example, the data in the second row and third column indicates that the gray-level co-occurrence pair (0,2) occurs 0 times, the first gray value in the gray-level co-occurrence pair (0,2) is 0, and the second gray value is 2.

[0078] Table 1

[0079]

[0080] It's important to note that there is no pixel overlap between corresponding pixels in any gray-level co-occurrence pair. For example, if the gray values ​​of the first and second pixels form a gray-level co-occurrence pair, then the gray values ​​of the third and fourth pixels form the next gray-level co-occurrence pair, not the second and third pixels. In other words, the pixels corresponding to the gray values ​​in two gray-level co-occurrence pairs do not overlap.

[0081] S31: Obtain the key sequence of each gray-level co-occurrence pair based on the statistical value of each gray-level co-occurrence pair in the gray-level co-occurrence matrix.

[0082] It should be noted that a key needs to be obtained first in order to encrypt the grayscale co-occurrence matrix.

[0083] Preferably, as an example, the key sequence of each gray-level co-occurrence pair is obtained based on the statistical value of each gray-level co-occurrence pair in the gray-level co-occurrence matrix, including:

[0084] The number of elements in the key sequence of each gray-level co-occurrence pair is calculated based on the occurrence frequency of each gray-level co-occurrence pair.

[0085] Generate some chaotic data and denote it as the second chaotic data. Take the vector formed by two second chaotic data as an element. Denote the sequence formed by L elements as the key sequence of each gray-level co-occurrence pair. L represents the number of elements in the key sequence of each gray-level co-occurrence pair. Denote the first second chaotic data of each element in the key sequence as the offset. Denote the second second chaotic data of each element in the key sequence as the span adjustment value.

[0086] For ease of explanation, the following example illustrates how to obtain the key sequence for each gray-level co-occurrence pair:

[0087] Assume the generated second chaotic data is -3, 10, 5, -8, and the gray-level co-occurrence pair is (40, 20). The vector [-3, 10] formed by the first chaotic data -3 and the second chaotic data 10 is used as one element; the vector [5, -8] formed by the third chaotic data 5 and the fourth chaotic data -8 is used as one element; the sequence of all elements is used as the key sequence {[-3, 10], [5, -8]} for the gray-level co-occurrence pair. In the key sequence, the first data (-3, 5) of all elements are offsets, and the second data (10, -8) of all elements are span adjustment values.

[0088] The above embodiments involve the second chaotic data and the number of elements in the key sequence of each gray-level co-occurrence pair. The method for obtaining the number of elements in the second chaotic data and the key sequence of each gray-level co-occurrence pair will be explained below.

[0089] First, we will introduce the method for obtaining the second chaotic data.

[0090] Preferably, as an example, the method for obtaining the second chaotic data includes:

[0091] The chaotic data generated using the Logistic mapping is denoted as the second chaotic data.

[0092] Then, the method for obtaining the number of elements in the key sequence of each gray-level co-occurrence pair is introduced.

[0093] Preferably, as an example, the number of elements in the key sequence of each gray-level co-occurrence pair includes:

[0094] The number of elements in the key sequence of each gray-level co-occurrence pair is obtained by multiplying the normalized value of the occurrence frequency of the gray-level co-occurrence pair by a preset adjustment value and then rounding it up.

[0095] S32: Add the mean of each gray-level co-occurrence pair to the offset value of each element in the key sequence in turn, add the span adjustment value of each element in the key sequence to the span adjustment value of each gray-level co-occurrence pair in turn to obtain the adjusted gray-level co-occurrence pair, and adjust the pixel value of each pixel in the final swapped face image based on the adjusted gray-level co-occurrence pair to obtain the encrypted face image.

[0096] Preferably, as an example, the mean of each gray-level co-occurrence pair is sequentially added to the offset value of each element in the key sequence, and the span value of each gray-level co-occurrence pair is sequentially added to the span adjustment value of each element in the key sequence to obtain the adjusted gray-level co-occurrence pair. Based on the adjusted gray-level co-occurrence pair, the pixel values ​​of each pixel in the finally swapped face image are adjusted to obtain the encrypted face image, including:

[0097] The absolute value of the difference between the two gray values ​​in a gray-level co-occurrence pair is taken as the span value of the gray-level co-occurrence pair;

[0098] The adjusted center of the gray-level co-occurrence pair is obtained by adding the mean of the two gray values ​​in the gray-level co-occurrence pair to the offset value of the first element in the key sequence. The adjusted span value is obtained by adding the span adjustment value of the first element to the span value of the gray-level co-occurrence pair. The adjusted first data of the gray-level co-occurrence pair is obtained by subtracting half of the adjusted span value from the adjusted center. The adjusted second data of the gray-level co-occurrence pair is obtained by adding half of the adjusted span value to the adjusted center. The data pair formed by the adjusted first data and the adjusted second data is used as the gray-level co-occurrence pair after the first round of encryption.

[0099] The adjusted gray-level co-occurrence pair is obtained by sequentially performing L rounds of encryption on each element in the key sequence.

[0100] The encrypted face image is obtained by replacing the gray values ​​of the corresponding pixels in each channel of the final swapped face image with the gray values ​​in the adjusted gray-level co-occurrence pairs.

[0101] For ease of explanation, the following example illustrates one round of encryption for gray-scale co-occurrence pairs:

[0102] The key sequence of the gray-scale co-occurrence pair (40,20) is {[-3,10],[5,-8]}, and the span value of the gray-scale co-occurrence pair is 20. The mean value of the gray-scale co-occurrence pair 30 is added to the first data of the first element of the key sequence -3 to obtain the adjusted center 27 of the gray-scale co-occurrence pair. Then, the span value 20 is added to the span adjustment value 10 to obtain the adjusted span value 30. The adjusted center 27 is subtracted from half of the adjusted span value 15 to obtain the adjusted first data 12 of the gray-scale co-occurrence pair. The adjusted center 27 is added to half of the adjusted span value 15 to obtain the adjusted second data 42 of the gray-scale co-occurrence pair. The data pair (12,42) formed by the adjusted first data 12 and the adjusted second data 42 is used as the gray-scale co-occurrence pair after the first round of encryption.

[0103] Using the encryption method described above, the adjusted gray-scale co-occurrence pair obtained by completing the second round of encryption using the second element in the key sequence is (26, 38).

[0104] Replace the first gray value 40 in the original gray-level co-occurrence pair with the first gray value 26 in the adjusted gray-level co-occurrence pair, and replace the second gray value 20 in the original gray-level co-occurrence pair with the second gray value 38 in the gray-level co-occurrence pair.

[0105] Understandably, the more times a gray-level co-occurrence pair appears, the more frequently the texture information feature appears, and the easier it is to identify. Therefore, this embodiment increases the difficulty of cracking by setting more encryption rounds for gray-level co-occurrence pairs that appear frequently.

[0106] Furthermore, by adding an offset value to the mean of the gray-level co-occurrence pair and an adjustment value to the original span value, the information of the original gray-level co-occurrence pair is masked, thereby effectively masking the texture information corresponding to the gray-level co-occurrence pair and effectively improving the encryption effect.

[0107] S33: To implement transmission processing.

[0108] Preferably, as an example, to implement the transmission processing, it includes:

[0109] The encrypted facial image is transmitted as the transmission object.

[0110] S4: Decrypt the encrypted face image.

[0111] Preferably, as an example, the decryption process for the encrypted face image includes:

[0112] Obtain the gray-level co-occurrence matrix of each channel of the encrypted face image according to the method in step S30. Calculate the number of elements in the key sequence based on the occurrence frequency of each gray-level co-occurrence pair in the gray-level co-occurrence matrix according to the method in step S31. Generate second chaotic data using Logistic mapping. Extract the key sequence of each gray-level co-occurrence pair from the second chaotic data based on the number of elements in the key sequence of each gray-level co-occurrence pair.

[0113] Based on the key sequence of each gray-level co-occurrence pair, the gray-level co-occurrence pair is decrypted by reversing the encryption method of each gray-level co-occurrence pair in step S32 to obtain the decrypted gray-level co-occurrence pair.

[0114] The image constructed using the grayscale values ​​of the decrypted grayscale co-occurrence pairs is denoted as the texture decryption image.

[0115] After the texture is decrypted, the image is uniformly divided into a first number of pre-examination image blocks to generate the first chaotic data. Based on the first chaotic data, the image block permutation process is performed in reverse order of the image block exchange method in step S21 to obtain the decrypted face image.

[0116] It should be noted that the hyperparameters of the Logistic mapping are predefined, so there is no need to pass the hyperparameters to the Logistic mapping. The decryptor can generate the corresponding first chaotic data and second chaotic data based on the Logistic mapping with the preset hyperparameters.

[0117] This invention also discloses a facial image transmission system for security gates, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a facial image transmission method for security gates according to the present invention.

[0118] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0119] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as resistive random access memory (DRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (DRAM), high-bandwidth memory, hybrid memory cube, etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device.

[0120] In the process of acquiring and using personal facial information, this invention strictly follows relevant laws and regulations to ensure personal privacy and personal information security. All acquisition of facial information has been with the explicit consent of the information subject or in accordance with other circumstances stipulated by laws and administrative regulations.

Claims

1. A method for transmitting facial images for use in security gates, characterized in that, Including the following steps: Acquire facial images; A face image is divided into several image blocks, and the image blocks are swapped several times until the degree of topological relationship destruction of the swapped face image is greater than a preset destruction threshold, thus obtaining the final swapped face image. The degree of topological relationship destruction reflects the difference between the positional relationship of feature points in the swapped face image and the positional relationship of human feature points in the original face image. The gray-level co-occurrence matrix of the final swapped face image is obtained. Based on the statistical values ​​of each gray-level co-occurrence pair in the gray-level co-occurrence matrix, a key sequence for each gray-level co-occurrence pair is obtained. Each element in the key sequence contains an offset value and a span adjustment value. The mean of each gray-level co-occurrence pair is sequentially added to the offset value of each element in the key sequence, and the span value of each gray-level co-occurrence pair is sequentially added to the span adjustment value of each element in the key sequence to obtain the adjusted gray-level co-occurrence pair. Based on the adjusted gray-level co-occurrence pair, the pixel values ​​of each pixel in the final swapped face image are adjusted to obtain the encrypted face image, thus achieving transmission processing.

2. The face image transmission method for security gates according to claim 1, characterized in that, The process of segmenting a face image into several image blocks includes: The face image is evenly divided into a preset first number of image blocks.

3. The method for transmitting facial images for security gates according to claim 1, characterized in that, The process of swapping image blocks in several rounds until the degree of topological disruption of the swapped face image exceeds a preset disruption threshold, resulting in the final swapped face image, includes: First chaotic data is generated. Image blocks are extracted based on a second preset number of the first chaotic data as active exchange image blocks, and image blocks are extracted based on a second preset number of the first chaotic data as passive exchange image blocks. The active and passive exchange image blocks are exchanged to obtain a face image after one round of exchange. The degree of topological relationship destruction is calculated based on the face image after one round of exchange. In response to the degree of topological relationship destruction not being greater than a preset destruction degree threshold, a second round of exchange is performed on each image block in the face image after one round of exchange. In response to the degree of topological relationship destruction being greater than the preset destruction degree threshold, the resulting exchanged face image is recorded as the final exchanged face image.

4. The face image transmission method for security gates according to claim 3, characterized in that, The calculation of the degree of topological relationship disruption based on the face images after one round of swapping includes: Obtain the corresponding feature points of each feature point in the face image after one round of swapping; In the face image, every two feature points are formed into a feature point pair. The corresponding feature point pair in the face image after one round of swapping is obtained. The vector formed by the two feature points in each feature point pair is recorded as the description vector. The degree of destruction of each feature point pair is obtained by multiplying the absolute value of the difference between the magnitude of each feature point pair and the description vector of the corresponding feature point pair by the included angle value. The normalized value of the mean of the degree of destruction of all feature point pairs is denoted as the degree of topological relationship destruction.

5. A method for transmitting facial images for security gates according to claim 1, characterized in that, The process of obtaining the gray-level co-occurrence matrix of the final swapped face image includes: In any direction, the gray values ​​of every two adjacent pixels in each channel of the final swapped face image are used to form a gray-level co-occurrence pair. The gray-level co-occurrence matrix is ​​obtained by statistically analyzing the gray-level co-occurrence pairs obtained in each channel image.

6. A method for transmitting facial images for security gates according to claim 5, characterized in that, The step of obtaining the key sequence for each gray-level co-occurrence pair based on the statistical values ​​of each gray-level co-occurrence pair in the gray-level co-occurrence matrix includes: The number of elements in the key sequence of each gray-level co-occurrence pair is calculated based on the occurrence frequency of each gray-level co-occurrence pair. Generate some chaotic data and denote it as the second chaotic data. Take the vector formed by two second chaotic data as an element. Denote the sequence formed by L elements as the key sequence of each gray-level co-occurrence pair, where L represents the number of elements in the key sequence of each gray-level co-occurrence pair. Denote the first second chaotic data of each element in the key sequence as the offset. Denote the second second chaotic data of each element in the key sequence as the span adjustment value.

7. A method for transmitting facial images for security gates according to claim 6, characterized in that, The step of calculating the number of elements in the key sequence of each gray-level co-occurrence pair based on the occurrence frequency of each gray-level co-occurrence pair includes: The number of elements in the key sequence of each gray-level co-occurrence pair is obtained by multiplying the normalized value of the occurrence frequency of the gray-level co-occurrence pair by a preset adjustment value and then rounding it up.

8. A method for transmitting facial images for security gates according to claim 6, characterized in that, The method for obtaining the adjusted gray-level co-occurrence pairs includes: The adjusted center of the gray-level co-occurrence pair is obtained by adding the mean of the two gray values ​​in the gray-level co-occurrence pair to the offset value of the first element in the key sequence. The adjusted span value is obtained by adding the span adjustment value of the first element to the span value of the gray-level co-occurrence pair. The adjusted first data of the gray-level co-occurrence pair is obtained by subtracting half of the adjusted span value from the adjusted center. The adjusted second data of the gray-level co-occurrence pair is obtained by adding half of the adjusted span value to the adjusted center. The data pair formed by the adjusted first data and the adjusted second data is used as the gray-level co-occurrence pair after the first round of encryption. The adjusted gray-level co-occurrence pair is obtained by performing L rounds of encryption on the gray-level co-occurrence pair using each element in the key sequence.

9. A method for transmitting facial images for security gates according to claim 5, characterized in that, The process of obtaining the encrypted face image by adjusting the pixel values ​​of each pixel in the final swapped face image based on the adjusted grayscale co-occurrence pair includes: The encrypted face image is obtained by replacing the gray values ​​of the corresponding pixels in each channel of the final swapped face image with the gray values ​​in the adjusted gray-level co-occurrence pairs.

10. A facial image transmission system for security gates, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a method for transmitting facial images for a security gate according to any one of claims 1-9.

Images