A processing system for multi-source sensitive data

Abstract

The application relates to the field of data processing, in particular to a processing system for multi-source sensitive data, which determines whether to perform frame screening analysis according to the content richness of a continuous frame sequence of a to-be-uploaded key frame; determines whether to replace the video frames in the continuous frame sequence of the to-be-uploaded key frame according to the encryption processing difficulty balance value of the continuous frame sequence of the to-be-uploaded key frame and target feature similarity; determines whether to perform local analysis on the to-be-uploaded key frame according to the image complexity of the to-be-uploaded key frame; determines whether to adjust the path length of the to-be-uploaded key frame or the iteration number of the focus area according to the regional complexity dispersion degree corresponding to the to-be-uploaded key frame; and determines to perform partition iteration encryption and global path encryption based on whether there is a focus area, so that the application can realize differential encryption processing of sensitive data.

Description

Technical Field

[0001] This invention relates to the field of data processing, and more particularly to a processing system for multi-source sensitive data. Background Technology

[0002] With the development of science and technology, the demand for internet data security is increasing daily. Video encryption technology, as a core means of ensuring the security of sensitive information within a video frame, can be used in various scenarios such as smart urban road monitoring, park security, vehicle sensing, face capture, and traffic recognition. Pixel position scrambling encryption is a common video encryption method in existing technologies. However, for the encryption of surveillance video transmitted over the internet, the encryption processing difficulty varies significantly across different surveillance frames. If a single encryption method is used for different surveillance frames, it can easily lead to low pixel scrambling levels in key frames and insufficient protection strength for sensitive targets. Therefore, how to achieve differentiated encryption processing of key surveillance frames and improve the real-time processing efficiency of encryption operations is a technical problem that urgently needs to be solved by those skilled in the art.

[0003] Chinese Patent Publication No. CN115297221B discloses an image encryption method based on pixel ring scrambling, comprising: constructing an encryption dictionary; scanning the image to be encrypted to obtain a pixel ring; obtaining the scrambling count and a first encryption dictionary; obtaining the pixels to be scrambled in the pixel ring; scrambling the pixel ring according to the first encryption dictionary and the scrambling count; and obtaining the ciphertext image based on the scrambled pixel ring. It is evident that the above technical solution has the following problems: when encrypting the image, it only uses a single pixel ring scrambling method. This encryption mechanism, which relies solely on position permutation, has limitations in encryption effectiveness for images with high complexity and rich texture details. Furthermore, it leads to a high risk of sensitive information leakage and poor real-time performance of overall encrypted data upload due to redundant processing of static background frames. Summary of the Invention

[0004] To address this, the present invention provides a processing system for multi-source sensitive data, which overcomes the problems of high risk of sensitive information leakage and poor real-time performance of overall encrypted data upload caused by the single and fixed encryption method in the prior art.

[0005] To achieve the above objectives, the present invention provides a processing system for multi-source sensitive data, comprising: The data analysis unit is used to determine whether to perform frame filtering analysis based on the content richness corresponding to the continuous frame sequence of the keyframe to be uploaded. The filtering and processing unit is used to perform frame filtering analysis. It determines whether to select video frames in the continuous frame sequence to replace the key frame to be uploaded based on the encryption processing difficulty balance value of the continuous frame sequence of the key frame to be uploaded, and the similarity of the target features. The complexity analysis unit is used to determine whether to perform local analysis on the keyframe to be uploaded based on the image complexity of the keyframe to be uploaded. The local analysis unit is used to perform local analysis, which determines the required path length for the keyframe to be uploaded or the required iteration number for the focused area based on the regional complexity dispersion of the keyframe to be uploaded. The encryption unit is used to determine whether to perform partitioned iterative encryption or global path encryption based on the existence of a focused region.

[0006] Furthermore, the data analysis unit responds to the condition that the content richness of the consecutive frame sequences of the keyframes to be uploaded is less than the preset content richness, and determines to perform frame filtering analysis.

[0007] Furthermore, in response to the condition that the encryption difficulty balance value of the continuous frame sequence of the key frame to be uploaded is less than or equal to the preset encryption difficulty balance value, the screening and processing unit determines to select video frames in the continuous frame sequence to replace the key frame to be uploaded, with encryption difficulty and target feature similarity as the constraint objectives.

[0008] Furthermore, the filtering and processing unit is also used to select video frames within a continuous frame sequence to replace the keyframes to be uploaded, based on the encryption processing difficulty and target feature similarity as constraints. This includes: Detect the similarity of target features among video frames in a continuous frame sequence and the difficulty of encryption processing; The evaluation value is determined based on the similarity of target features and the difficulty of encryption processing; The video frame with the highest evaluation value will be selected to replace the keyframe to be uploaded.

[0009] Furthermore, the complexity analysis unit responds to the condition that the image complexity of the keyframe to be uploaded is less than or equal to the preset image complexity, and determines to perform local analysis on the keyframe to be uploaded.

[0010] Furthermore, the local analysis unit performs local analysis on the keyframe to be uploaded, including: Detect the regional complexity dispersion of the keyframe to be uploaded; For conditions where the regional complexity dispersion is greater than the preset regional complexity dispersion, the required path length for the keyframe to be uploaded is adjusted accordingly. For conditions where the regional complexity dispersion is less than or equal to the preset regional complexity dispersion, the number of iterations for the focused region is adjusted.

[0011] Furthermore, the required path length for the keyframes to be uploaded is adjusted to be increased. The increase in the length of the demand path is negatively correlated with the image complexity.

[0012] Furthermore, the adjustment to increase the number of iterations for the focus area's needs is implemented; The increase in the number of iterations required for the focused region is positively correlated with the texture complexity corresponding to the focused region.

[0013] Furthermore, the encryption unit responds to the condition that a focused region exists and determines to perform partitioned iterative encryption. In the partitioned iterative encryption, each sub-region of the keyframe to be uploaded is recorded as the target region to be processed, and image encryption processing is performed on each target region to be processed. If the encryption unit responds to the condition that there is no focal region, it determines to perform global path encryption. In global path encryption, the entire keyframe to be uploaded is recorded as the target area to be processed, and image encryption processing is performed on the target area to be processed.

[0014] Furthermore, image encryption processing includes: Starting from a selected starting point, a pixel block within the target processing area is used to move and access the pixel blocks according to the required path length based on preset movement conditions, and the movement path is recorded in matrix form. Move the pixel visited for the first time in the original image of the target area to be processed to the position visited for the (i+1)th time in the path, and so on. After all pixels have been moved, the scrambled image is obtained, which is recorded as the completion of a single iteration. Continue until the number of iterations meets the required number of iterations.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: In the technical solution of the present invention, the key frame screening analysis is adaptively determined according to the content richness of the continuous frame sequence to achieve accurate selection of key frames, improve the adaptability of subsequent task processing, and adjust the encryption method of key frames according to the image complexity and regional complexity dispersion of key frames, thereby improving the encryption protection strength of sensitive information, solving the defect that a single encryption method is easy to reverse restore, and further meeting the needs of efficient and secure processing of sensitive data from multiple sources of video surveillance.

[0016] Furthermore, this invention determines whether to perform keyframe screening based on the content richness of consecutive frames, and selects keyframes to be uploaded based on the balance value of encryption processing difficulty and the similarity of target features. This reduces the computational overhead of redundant static background frames. Moreover, it adaptively adjusts the encryption path length and iteration number based on image complexity and regional complexity dispersion, solving the problems of wasted computing power or insufficient protection of sensitive data caused by fixed encryption methods, and further improving the real-time processing efficiency of sensitive data in surveillance videos.

[0017] Furthermore, this invention determines whether to perform local analysis based on image complexity, and adjusts the encryption path length or the number of regional iterations based on the regional complexity dispersion, so as to realize the encryption method dynamically adapts to the features of key frames, thus balancing the reduction of computing power redundancy for low-texture images and the enhancement of encryption strength for high-texture-sensitive images.

[0018] Furthermore, the present invention adopts a pixel block traversal path iterative scrambling encryption method, which moves pixels according to a preset path and performs multiple rounds of iteration to complete image scrambling, breaks the spatial correlation between pixels, overcomes the problem of insufficient obfuscation degree of a single encryption method and the problem that sensitive images are easy to reverse engineer, and further improves the reliability of encryption protection for sensitive information. Attached Figure Description

[0019] Figure 1 This is a unit connection diagram of the multi-source sensitive data processing system according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating how to determine whether to perform frame filtering analysis based on content richness, as described in an embodiment of the present invention. Figure 3 This is a flowchart illustrating the process of determining whether to replace the keyframe to be uploaded based on the encryption difficulty balance value in an embodiment of the present invention. Figure 4 This is a flowchart illustrating the adjustment of the encryption method based on the regional complexity and dispersion in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0021] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0023] Please see Figure 1 The diagram shown is a unit connection diagram of a multi-source sensitive data processing system according to an embodiment of the present invention. The system includes: The data analysis unit is used to determine whether to perform frame filtering analysis based on the content richness corresponding to the continuous frame sequence of the keyframe to be uploaded. The filtering and processing unit is used to perform frame filtering analysis. It determines whether to select video frames in the continuous frame sequence to replace the key frame to be uploaded based on the encryption processing difficulty balance value of the continuous frame sequence of the key frame to be uploaded, and the similarity of the target features. The complexity analysis unit is used to determine whether to perform local analysis on the keyframe to be uploaded based on the image complexity of the keyframe to be uploaded. The local analysis unit is used to perform local analysis, which determines the required path length for the keyframe to be uploaded or the required iteration number for the focused area based on the regional complexity dispersion of the keyframe to be uploaded. The encryption unit is used to determine whether to perform partitioned iterative encryption or global path encryption based on the existence of a focused region. Please see Figure 2 As shown, this is a flowchart of an embodiment of the present invention for determining whether to perform frame filtering analysis based on content richness. The data analysis unit responds to the condition that the content richness of the continuous frame sequence of the key frame to be uploaded is less than the preset content richness and determines to perform frame filtering analysis.

[0024] In this embodiment of the invention, the data analysis unit includes a richness acquisition module for obtaining content richness. The content richness acquisition method includes determining content richness based on vehicle position change and background similarity. The content richness is positively correlated with vehicle position change and negatively correlated with background similarity. In this embodiment, scaling processing is performed on the position offset and background similarity respectively. The difference between the scaled vehicle position change and the scaled background similarity is recorded as the content richness. The scaling processing of vehicle position change and background similarity involves multiplying each by a scaling factor. Each has an independent scaling factor, and the sum of these scaling factors is 1. A larger scaling factor indicates greater attention to the data item multiplied by the scaling factor, and a greater impact of the data item on the content richness. This is easily understood by those skilled in the art and will not be elaborated further. It is understandable that the lower the content richness, the weaker the content difference between video frames in a continuous frame sequence, and the more similar the picture content between the frames. Therefore, frame screening analysis is performed to select video frames that are highly similar to the key frame to be uploaded and have lower encryption difficulty as replacement frames, and then encryption processing is performed.

[0025] In this embodiment of the invention, the preset content richness setting method is as follows: prepare several monitoring video samples, perform frame screening analysis and judgment on each sample with different content richness candidate thresholds, and record the content richness candidate threshold corresponding to the test result with the highest consistency between the frame screening analysis judgment result and the manual screening result as the preset content richness. The manual screening result is determined by the user based on human experience to determine whether to perform frame screening analysis.

[0026] In this embodiment of the invention, the process by which the data analysis unit obtains the vehicle position change degree includes: performing vehicle detection on a continuous frame sequence of keyframes to be uploaded based on a target detection model, obtaining vehicle regions in each video frame, outputting bounding boxes for each vehicle in each video frame, and for each pair of adjacent video frames, obtaining the coordinates of the center point of the bounding box of each vehicle in the previous frame and the coordinates of the center point of the bounding box of each vehicle in the next frame, calculating the Euclidean distance between the coordinates of the center point of the previous frame and the center point of the next frame, denoted as the position offset, and recording the average value of all obtained position offsets as the vehicle position change degree. It can be understood that the vehicle position change degree is used to reflect the degree of drastic positional change of the target of interest within a video frame; the greater the vehicle position change degree, the greater the difference in the spatial characteristics of the target of interest between video frames.

[0027] In this embodiment of the invention, the process of obtaining background similarity by the data analysis unit includes: performing background separation on video frames, determining the area in each video frame other than the vehicle area as the background area, obtaining the background areas of each pair of adjacent video frames for each pair of adjacent video frames, obtaining color histogram features by statistically analyzing the pixel values ​​of each pixel in the background area, distributing the pixel values ​​of each pixel in the background area across each color channel according to a preset grayscale range to generate a color histogram vector, calculating the cosine similarity between the color histogram vectors of the background areas of the two video frames, and recording the average of all obtained cosine similarities as the background similarity. It can be understood that background similarity is used to characterize the degree of similarity between background areas within each video frame; the greater the background similarity, the smaller the scene changes between video frames.

[0028] Specifically, if the content richness of a consecutive frame sequence of a keyframe to be uploaded is greater than or equal to a preset content richness, it is determined that there is no need to perform frame filtering analysis, and the keyframe to be uploaded is directly selected for complexity analysis.

[0029] Please see Figure 3As shown, it is a flowchart of the analysis of whether to replace the key frame to be uploaded based on the encryption processing difficulty balance value in an embodiment of the present invention. The filtering processing unit responds to the condition that the encryption processing difficulty balance value of the continuous frame sequence of the key frame to be uploaded is less than or equal to the preset encryption processing difficulty balance value, and determines to select video frames in the continuous frame sequence to replace the key frame to be uploaded with encryption processing difficulty and target feature similarity as the constraint target.

[0030] In this embodiment of the invention, the filtering processing unit includes an equalization value acquisition module for obtaining an encryption processing difficulty equalization value. The method for obtaining the encryption processing difficulty equalization value includes: determining it based on the degree of difference in encryption processing difficulty among video frames within a continuous frame sequence of each keyframe to be uploaded; the greater the degree of difference in encryption processing difficulty among video frames, the smaller the encryption processing difficulty equalization value. In this embodiment, the reciprocal of the variance of the encryption processing difficulty corresponding to each video frame is obtained and denoted as the encryption processing difficulty equalization value. It is understood that... In this embodiment of the invention, the filtering processing unit includes a difficulty acquisition module for obtaining the encryption processing difficulty. The method for obtaining the encryption processing difficulty includes determining the encryption processing difficulty based on the number of targets of interest within a video frame and the spatial depth difference value of the targets of interest. The encryption processing difficulty is negatively correlated with both the number of targets of interest and the spatial depth difference value of the targets of interest. In this embodiment, the number of targets of interest and the spatial depth difference value of the targets are scaled respectively, and the reciprocal of the sum of the scaled number of targets of interest and the spatial depth difference value of the targets is denoted as the encryption processing difficulty.

[0031] In this embodiment of the invention, the process by which the filtering processing unit obtains the number of targets of interest includes: calling a pre-trained target detection model to detect the number of targets of interest in the keyframes to be uploaded, and outputting the number of targets of interest in each keyframe to be uploaded. The target detection model includes, but is not limited to, YOLO series models.

[0032] In this embodiment of the invention, the process by which the filtering processing unit obtains the spatial depth difference value of the target of interest includes: obtaining the height and width of the bounding box of each target of interest, calculating the area of ​​the bounding box, and recording the area of ​​the bounding box of each target of interest as the depth representation value of that target. The variance of each depth representation value is calculated, and the result is recorded as the spatial depth difference value.

[0033] In this embodiment of the invention, the method for obtaining the encryption processing difficulty balance value includes: obtaining the encryption processing difficulty of each video frame, analyzing the balance between encryption processing difficulties, and obtaining the encryption processing difficulty balance value.

[0034] In this embodiment, the inverse of the variance of the encryption difficulty of video frames is denoted as the encryption difficulty balance value. It can be understood that the encryption difficulty balance value reflects the degree of difference in encryption difficulty between video frames in a continuous frame sequence. The smaller the encryption difficulty balance value, the more significant the difference in encryption difficulty between video frames, thus facilitating the replacement of keyframes to be uploaded.

[0035] In this embodiment of the invention, the preset encryption processing difficulty balance value is set as follows: several video samples containing continuous frame sequences are prepared, and the continuous frame sequences in each sample are subjected to frame screening analysis with different candidate thresholds for encryption processing difficulty balance value. The candidate threshold for encryption processing difficulty balance value corresponding to the test result with the smallest ratio of the encryption processing difficulty of the replacement frame to the encryption processing difficulty of the original key frame to be uploaded after frame screening analysis is recorded as the preset encryption processing difficulty balance value.

[0036] In this embodiment of the invention, the method for obtaining target feature similarity includes: determining target feature similarity based on the number of matching feature point pairs, wherein the number of matching feature point pairs is positively correlated with target feature similarity. In this embodiment, the local feature point set of the keyframe to be uploaded and the local feature point set of the selected video frame are obtained respectively. The distance between each feature point descriptor in the keyframe to be uploaded and each feature point descriptor in the selected video frame is calculated. Feature point pairs with a distance less than a preset threshold are recorded as matching feature point pairs. The ratio of the number of matching feature point pairs to the total number of feature points in the keyframe to be uploaded is recorded as the target feature similarity. The local feature points are image keypoints in the video frame that have salientity and uniqueness, including but not limited to corner points, edge points, and spots.

[0037] Specifically, the filtering and processing unit is also used to select video frames within a continuous frame sequence to replace the keyframes to be uploaded, based on the constraints of encryption processing difficulty and target feature similarity, including: Detect the similarity of target features among video frames in a continuous frame sequence and the difficulty of encryption processing; The evaluation value is determined based on the similarity of target features and the difficulty of encryption processing; The video frame with the highest evaluation value will be selected to replace the keyframe to be uploaded.

[0038] In this embodiment of the invention, the keyframe replacement process performed by the filtering processing unit includes: obtaining a continuous frame sequence corresponding to the keyframe to be uploaded; determining the encryption processing difficulty of each video frame in the continuous frame sequence; determining an encryption processing difficulty balance value based on the encryption processing difficulty of all video frames; comparing the encryption processing difficulty balance value with a preset encryption processing difficulty balance value; if the encryption processing difficulty balance value is less than or equal to the preset encryption processing difficulty balance value, determining that video frames within the continuous frame sequence are selected to replace the keyframe to be uploaded, with encryption processing difficulty and target feature similarity as constraints; if the encryption processing difficulty balance value is greater than the preset encryption processing difficulty balance value, determining that frame filtering analysis is unnecessary and retaining the original keyframe to be uploaded.

[0039] The process of replacing uploaded keyframes includes: after determining the replacement frame in frame filtering analysis, extracting the video frame data of the replacement frame from the continuous frame sequence, establishing a mapping relationship between the original keyframe to be uploaded and the replacement frame, recording the mapping relationship in a replacement mapping table, inputting the video frame data of the replacement frame into an encryption unit, and having the encryption unit perform encryption processing on the replacement frame according to the aforementioned image encryption processing method. After completing the switching of the encryption processing object, the system performs buffering or discarding processing on the original keyframe to be uploaded. The replacement mapping table includes at least the frame identifier of the original keyframe to be uploaded, the frame identifier of the replacement frame, the position information of the replacement frame in the continuous frame sequence, and the timestamp of the replacement operation.

[0040] Specifically, the complexity analysis unit responds to the condition that the image complexity of the keyframe to be uploaded is less than or equal to the preset image complexity, and determines to perform local analysis on the keyframe to be uploaded.

[0041] In this embodiment of the invention, the complexity analysis unit includes a complexity acquisition module for acquiring image complexity. The image complexity acquisition method includes determining image complexity based on the number of edge pixels, where the number of edge pixels in the video frame is positively correlated with image complexity. In this embodiment, an edge detection operator is used to perform edge detection on the video frame, the number of detected edge pixels is counted, and the ratio of the number of edge pixels to the total number of pixels in the video frame is calculated. This ratio is recorded as the image complexity. The edge detection operator includes, but is not limited to, the Sobel operator, the Canny operator, and the Prewitt operator. It can be understood that a larger number of edge pixels indicates richer object contours and details in the image, more complex image content, and thus greater image complexity; conversely, a smaller number of edge pixels indicates fewer object contours and details in the image, simpler image content, and thus lower image complexity.

[0042] In this embodiment of the invention, the preset image complexity is set as follows: several keyframe samples containing focused and non-focused regions are prepared, and local analysis is performed on each sample with different image complexity candidate thresholds. The image complexity candidate threshold corresponding to the test result with the highest consistency rate between the local analysis result and the manual annotation result is selected as the preset image complexity.

[0043] Specifically, for keyframes to be uploaded whose image complexity exceeds the preset image complexity, there is no need to perform local analysis on the keyframes to be uploaded; instead, global path encryption is used.

[0044] Please see Figure 4 The flowchart shown is a process for determining the encryption method adjustment based on the regional complexity dispersion in an embodiment of the present invention. The local analysis unit performs local analysis on the key frame to be uploaded, including: Detect the regional complexity dispersion of the keyframe to be uploaded; For conditions where the regional complexity dispersion is greater than the preset regional complexity dispersion, the required path length for the keyframe to be uploaded is adjusted accordingly. For conditions where the regional complexity dispersion is less than or equal to the preset regional complexity dispersion, the number of iterations for the focused region is adjusted.

[0045] In this embodiment of the invention, the local analysis unit includes a local analysis module for obtaining the regional complexity dispersion. The method for obtaining the regional complexity dispersion includes: dividing the keyframe to be uploaded into several sub-regions of the same size; for each sub-region, obtaining the image complexity corresponding to that sub-region. The method for determining the image complexity of the sub-region is the same as the method for obtaining the image complexity, including but not limited to methods based on image information entropy, edge density, and pixel gray value variance. After determining the image complexity of each sub-region, the dispersion of the image complexity of all sub-regions is calculated, and the obtained dispersion is recorded as the regional complexity dispersion, wherein the regional complexity dispersion is the standard deviation of the image complexity of all sub-regions.

[0046] In this embodiment, the preset regional complexity dispersion is set as follows: several key frame samples containing focused and non-focused regions are prepared, and local analysis and encryption methods are adjusted for each sample using different regional complexity dispersion candidate thresholds. The regional complexity dispersion candidate threshold corresponding to the test result with the largest degree of disorder in the focused region after encryption of the key frame to be uploaded is selected as the preset regional complexity dispersion.

[0047] Specifically, the adjustment for the required path length of the keyframe to be uploaded is to increase it; the increase in the required path length is negatively correlated with the image complexity.

[0048] In this embodiment of the invention, the required path length is the number of steps included in the pixel block movement access path during image encryption processing, i.e., the number of pixel blocks accessed in the knight's patrol path. The larger the required path length, the more pixel blocks participate in scrambling in a single iteration, the wider the range of pixel positions being scrambled, and the higher the encryption strength. The increase in the required path length is negatively correlated with the image complexity of the keyframe to be uploaded: when the image complexity is high, the image itself already has rich texture information, and the spatial correlation between pixels is weak, so a small increase in path length is sufficient to obtain a sufficient scrambling effect; when the image complexity is low, the image texture is sparse, and the spatial correlation between pixels is strong, requiring a larger increase in path length to break the regular distribution between pixels to ensure the degree of confusion after scrambling. In this embodiment, the adjusted required path length is the product of the unadjusted required path length and the adjustment coefficient, where the adjustment coefficient is the ratio of the preset image complexity to the image complexity. Understandably, for keyframes with high image complexity, their textures already possess a high degree of visual obfuscation, so there's no need to excessively increase the path length to ensure encryption effectiveness. For keyframes with low image complexity, the path length needs to be increased to compensate for the weak scrambling effect caused by insufficient texture information. The greater the user's sensitivity to the encryption strength of low-texture images, the greater the initial value of the required path length.

[0049] Specifically, the adjustment for the number of iterations required for the focused area is to increase the adjustment; The increase in the number of iterations required for the focused region is positively correlated with the texture complexity corresponding to the focused region.

[0050] In this embodiment of the invention, the process of adjusting the number of iterations required for the focused region includes: In this embodiment of the invention, the adjustment of the required iteration number for the focused area is an increase. The required iteration number refers to the number of times pixel scrambling and diffusion operations are repeatedly performed on the same image to be processed during image encryption processing. The larger the required iteration number, the more rounds the pixel positions are rearranged and the pixel values ​​are diffused, resulting in higher encryption strength. The increase in the required iteration number for the focused area is positively correlated with the texture complexity of the focused area: when the texture complexity of the focused area is high, the area contains rich detailed information, requiring higher encryption strength and necessitating a larger increase in the iteration number to ensure the security of sensitive information; when the texture complexity of the focused area is low, a smaller increase is sufficient to meet the encryption requirements. In this embodiment, the texture complexity of the focused area is determined based on the density of edge pixels within the area; a higher edge pixel density results in higher texture complexity. It is understood that the focused area is the carrier of sensitive information in the monitoring image, and its encryption strength should be higher than that of non-focused areas. Increasing the required iteration number can further improve the scrambling and diffusion effect of the focused area, effectively resisting attacks and reverse engineering based on statistical features. The greater the user's sensitivity to the encryption strength of sensitive areas, the greater the increase in the required iteration number for the focused area.

[0051] In this embodiment of the invention, the increase in the required iteration number of the focused area is determined through simulation: focused area samples with different texture complexities are collected, and each sample is encrypted with different increase rates. The information entropy of the encrypted focused area and the correlation coefficient between adjacent pixels are used as evaluation indicators. The increase rate at which the information entropy improvement meets user requirements and the correlation coefficient between adjacent pixels decreases to below a preset threshold is recorded as the preset increase rate. The greater the user's sensitivity to the encryption strength of the focused area, the larger the value of the preset increase rate.

[0052] Specifically, the encryption unit responds to the condition that there is a focused area and determines to perform partitioned iterative encryption. In the partitioned iterative encryption, each sub-region of the key frame to be uploaded is recorded as the target to be processed region, and image encryption processing is performed on each target to be processed region. If the encryption unit responds to the condition that there is no focal region, it determines to perform global path encryption. In global path encryption, the entire keyframe to be uploaded is recorded as the target area to be processed, and image encryption processing is performed on the target area to be processed.

[0053] In this embodiment of the invention, the focused region is an image sub-region in the keyframe to be uploaded that meets a preset focusing condition. The preset focusing condition is determined based on a comparison between the image complexity of the sub-region and a preset complexity threshold. Specifically, the keyframe to be uploaded is divided into several sub-regions of the same size, and the image complexity of each sub-region is calculated. When the image complexity of a sub-region is greater than the preset complexity threshold, the sub-region is marked as a focused region. It can be understood that the focused region is a region in the keyframe to be uploaded with rich texture details and high image complexity. The focused region requires higher encryption strength than the non-focused region. Therefore, the system implements a differentiated encryption strategy by identifying the focused region. The greater the user's sensitivity to the accuracy of sensitive target identification, the smaller the value of the preset complexity threshold.

[0054] Specifically, image encryption processing includes: Starting from any pixel block within the target processing area, the system moves and accesses the pixel blocks according to the required path length based on preset movement conditions, and records the path in matrix form. Move the pixel visited for the first time in the original image of the target area to be processed to the position visited for the (i+1)th time in the path, and so on. After all pixels have been moved, the scrambled image is obtained, which is recorded as the completion of a single iteration. Continue until the number of iterations meets the required number of iterations.

[0055] In this embodiment of the invention, the process of encrypting the keyframe to be uploaded by the encryption unit includes: acquiring an image to be processed, wherein the image to be processed is the keyframe to be uploaded; converting the image to be processed into a pixel matrix, wherein the number of rows in the pixel matrix is ​​equal to the height of the image to be processed in pixels, the number of columns in the pixel matrix is ​​equal to the width of the image to be processed in pixels, and the value of each element in the pixel matrix corresponds to the grayscale value or color component value of the pixel at that position; inputting a preset chaotic initial value into a chaotic mapping system, which generates a chaotic sequence; the chaotic mapping system is highly sensitive to the initial conditions, and different initial values ​​will produce completely different chaotic sequences; using the generated chaotic sequence as a randomization control parameter for path generation, the starting point position of the knight's patrol and the direction of each step are dynamically determined through the chaotic sequence. Based on the generated knight's patrol path, pixel scrambling is performed on the image to be processed. Specifically, the first visited pixel in the original image is moved to the second visited position in the path; the second visited pixel in the original image is moved to the third visited position in the path; and so on; the (i-1)th visited pixel in the original image is moved to the ith visited position in the path; and the ith visited pixel in the original image is moved to the first visited position in the path, where i is the required path length. After all the pixel movements are completed, a scrambled image is obtained. The pixel values ​​of each pixel in the scrambled image are then XORed sequentially with the chaotic sequence according to the patrol path order. This XOR operation allows any change in pixel value to propagate to multiple subsequent pixels, further enhancing the encryption effect. The encrypted image obtained after a single scrambling and propagation operation is recorded as a single iteration completed. Determine if the current iteration count has reached the required iteration count. If the current iteration count is less than the required iteration count, use the current encrypted image as the new image to be processed, and re-execute the knight's patrol path generation, pixel scrambling operation, and pixel value diffusion operation to proceed to the next iteration. If the current iteration count has reached the required iteration count, stop the iteration and output the current encrypted image as the encrypted image.

[0056] In this embodiment of the invention, the specific process for decrypting keyframes includes: obtaining a decryption key and performing a key validity verification step. The key validity verification includes: verifying whether each parameter in the decryption key is within a valid value range, verifying whether the integrity of the decryption key meets the requirements, and verifying whether the association identifier between the decryption key and the image to be decrypted matches. It is understood that if the decryption key fails the validity verification, the system terminates the decryption process and outputs a decryption failure message.

[0057] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0058] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A processing system for multi-source sensitive data, characterized in that, include: The data analysis unit is used to determine whether to perform frame filtering analysis based on the content richness corresponding to the continuous frame sequence of the keyframe to be uploaded. The filtering and processing unit is used to perform frame filtering analysis. It determines whether to select video frames in the continuous frame sequence to replace the key frame to be uploaded based on the encryption processing difficulty balance value of the continuous frame sequence of the key frame to be uploaded, and the similarity of the target features. The complexity analysis unit is used to determine whether to perform local analysis on the keyframe to be uploaded based on the image complexity of the keyframe to be uploaded. The local analysis unit is used to perform local analysis, which determines the required path length for the keyframe to be uploaded or the required iteration number for the focused area based on the regional complexity dispersion of the keyframe to be uploaded. The encryption unit is used to determine whether to perform partitioned iterative encryption or global path encryption based on the existence of a focused region.

2. The processing system for multi-source sensitive data according to claim 1, characterized in that, The data analysis unit responds to the condition that the content richness of the consecutive frame sequences of the keyframes to be uploaded is less than the preset content richness, and then determines to perform frame filtering analysis.

3. The processing system for multi-source sensitive data according to claim 2, characterized in that, The filtering and processing unit responds to the condition that the encryption processing difficulty balance value of the continuous frame sequence of the key frame to be uploaded is less than or equal to the preset encryption processing difficulty balance value, and determines to select video frames in the continuous frame sequence to replace the key frame to be uploaded, with encryption processing difficulty and target feature similarity as the constraint objectives.

4. The processing system for multi-source sensitive data according to claim 3, characterized in that, The filtering and processing unit is also used to select video frames within a continuous frame sequence to replace the keyframes to be uploaded, based on the encryption difficulty and target feature similarity as constraints. Detect the similarity of target features among video frames in a continuous frame sequence and the difficulty of encryption processing; The evaluation value is determined based on the similarity of target features and the difficulty of encryption processing; The video frame with the highest evaluation value will be selected to replace the keyframe to be uploaded.

5. The processing system for multi-source sensitive data according to claim 3, characterized in that, The complexity analysis unit responds to the condition that the image complexity of the keyframe to be uploaded is less than or equal to the preset image complexity, and determines to perform local analysis on the keyframe to be uploaded.

6. The processing system for multi-source sensitive data according to claim 5, characterized in that, The local analysis unit performs local analysis on the keyframe to be uploaded, including: Detect the regional complexity dispersion of the keyframe to be uploaded; For conditions where the regional complexity dispersion is greater than the preset regional complexity dispersion, the required path length for the keyframe to be uploaded is adjusted accordingly. For conditions where the regional complexity dispersion is less than or equal to the preset regional complexity dispersion, the number of iterations for the focused region is adjusted.

7. The processing system for multi-source sensitive data according to claim 6, characterized in that, The required path length for the keyframes to be uploaded is adjusted to be increased. The increase in the length of the demand path is negatively correlated with the image complexity.

8. The processing system for multi-source sensitive data according to claim 6, characterized in that, The adjustment to increase the number of iterations for the focus area is now implemented. The increase in the number of iterations required for the focused region is positively correlated with the texture complexity corresponding to the focused region.

9. The processing system for multi-source sensitive data according to claim 6, characterized in that, The encryption unit responds to the condition that a focused area exists and determines to perform partitioned iterative encryption. In the partitioned iterative encryption, each sub-region of the keyframe to be uploaded is recorded as the target to be processed area, and image encryption processing is performed on each target to be processed area. If the encryption unit responds to the condition that there is no focal region, it determines to perform global path encryption. In global path encryption, the entire keyframe to be uploaded is recorded as the target area to be processed, and image encryption processing is performed on the target area to be processed.

10. The processing system for multi-source sensitive data according to claim 4, characterized in that, Image encryption processing includes: Starting from a selected starting point, a pixel block within the target processing area is used to move and access the pixel blocks according to the required path length based on preset movement conditions, and the movement path is recorded in matrix form. Move the pixel visited for the first time in the original image of the target area to be processed to the position visited for the (i+1)th time in the path, and so on. After all pixels have been moved, the scrambled image is obtained, which is recorded as the completion of a single iteration. Continue until the number of iterations meets the required number of iterations.

Images