Privacy encryption method and system based on palm vein features

Abstract

The invention discloses a palm vein privacy encryption recognition method and system and a storage medium, and the method comprises the steps: obtaining an original palm vein image, and carrying out the image correction, and obtaining a balanced correction image; key feature points are extracted, a density distribution diagram is constructed, and a structured feature data set is obtained through spatial structure recombination; performing protective coding on the structured data set to generate an encrypted feature combination; if the visibility is insufficient, adjusting a matching error tolerance interval to obtain a fine adjustment feature set; and comparing the fine adjustment feature set with a pre-established palm vein template library, and outputting an identity verification result. In the model training stage, historical palm vein data is preprocessed and dimensionally reduced, and an encrypted feature model is constructed and optimized based on a particle swarm algorithm. According to the invention, high-precision and high-safety palm vein real-time identity verification can be realized.

Description

Technical Field

[0001] This invention relates to the field of biometric identification, and in particular to a privacy encryption method and system based on palm vein features. Background Technology

[0002] In the field of biometric identification, the importance of privacy protection technology is increasingly prominent, as it is directly related to the security of personal identity information and the cornerstone of social trust. With the acceleration of digitalization, especially the effective protection of user privacy while ensuring identification accuracy, has become a crucial issue that cannot be ignored in this field. As unique identifiers, biometrics have particularly urgent security requirements; any leakage could have irreversible consequences.

[0003] Current palm vein recognition technology mainly involves acquiring palm vein images using infrared imaging equipment. After image preprocessing, key information is extracted using the unique characteristics of the vein texture. The extracted features are then combined with an encryption algorithm to generate an encryption key or directly used for data encryption. During identity authentication, the palm vein image is acquired again, and the features are compared with the encrypted information to complete identity verification.

[0004] Existing palm vein recognition solutions cannot dynamically adjust acquisition parameters when ambient light, skin reflection, and saturation change, resulting in large fluctuations in image quality. Privacy protection relies solely on traditional encryption and does not encrypt the vein topology, making it vulnerable to reconstruction after leakage. Therefore, existing technologies suffer from issues of low recognition accuracy and security. Summary of the Invention

[0005] This invention provides a privacy encryption method and system based on palm vein features to solve the problem of low accuracy and security in existing technologies for palm vein recognition.

[0006] In a first aspect, to address the aforementioned technical problems, the present invention provides a privacy encryption method and system based on palm vein features, comprising: The original palm vein image is acquired and preprocessed to obtain the initial corrected image; The initial corrected image is segmented and detected to mark regions with uneven saturation and generate a saturation distribution map. Based on the saturation distribution map and the local brightness contrast coefficient, the parameters of the uneven saturation region are adjusted to generate a balanced correction image. Key palm vein feature points are extracted from the balanced and corrected image, a feature point density distribution map is constructed, and spatial structure is reorganized to obtain a structured feature dataset. The geometric topological relationships of feature points are extracted from the structured feature dataset and protected by encoding to generate encrypted feature combinations; If the visibility of the feature point density distribution map is lower than the preset visibility threshold for the encrypted feature combination, the matching error tolerance interval is adjusted to obtain a fine-tuned feature set; The fine-tuned feature set is compared with a pre-established palm vein template library to generate a palm vein authentication result.

[0007] In one optional implementation, the acquisition of raw palm vein image data and preprocessing to obtain an initial corrected image includes: Acquire the original palm vein image, and when the light influence factor exceeds the preset influence range, perform light compensation on the original palm vein image to obtain the first correction data; The reflectance of the first correction data is adjusted using a preset correction coefficient table to determine the second correction data; The grayscale distribution of the second correction data is adjusted and the contrast is enhanced to obtain the initial correction image.

[0008] In one optional implementation, the step of performing region segmentation and detection on the initial corrected image, marking regions with uneven saturation, and generating a saturation distribution map includes: The initial corrected image is processed in layers to obtain the saturation fluctuation threshold range corresponding to the data fluctuation of each layer, and the saturation fluctuation reference value of each layer is determined. By combining the saturation fluctuation benchmark value with the region segmentation grid density, block processing is performed to obtain multiple regions; The regions in the multiple regions whose saturation fluctuations exceed a preset fluctuation threshold are marked as the saturation unevenness regions, thus obtaining a set of marked regions; The marked region set is scanned to obtain saturation distribution data and thus saturation distribution characteristics. The saturation distribution features are integrated to obtain a saturation distribution mapping map.

[0009] In one optional implementation, the step of adjusting the parameters of the uneven saturation region based on the saturation distribution map and the local brightness contrast coefficient to generate a balanced corrected image includes: Extract the specific locations of the uneven saturation regions in the saturation distribution map; The specific locations of the uneven areas are initially marked, and the marked areas that exceed the preset deviation threshold are determined as the target adjustment areas; Based on the location data of the target adjustment area, the corresponding local brightness distribution information is obtained and quantified. The target adjustment area is divided into multiple target adjustment sub-regions, and the contrast coefficient value is obtained by continuously integrating each target adjustment sub-region. Adjust the corresponding correction parameters for the contrast coefficient value of the target adjustment sub-region, increase the grayscale stretching amplitude of the target adjustment sub-region with a contrast coefficient lower than the preset contrast threshold, and determine the adjusted parameter configuration; Based on the adjusted parameter configuration, and combined with the overall saturation and brightness distribution data, the equalization correction image is generated.

[0010] In one optional implementation, the step of extracting key palm vein feature points from the equalization-corrected image, constructing a feature point density distribution map, and performing spatial structure reconstruction to obtain a structured feature dataset includes: The original point data is obtained from the equalization and correction image, and edge detection and noise filtering are performed to obtain an initial set of feature points; A density distribution map is constructed based on the initial set of feature points, and the density distribution map is divided into multiple density sub-regions. The distribution density of each sub-region is calculated to determine the density distribution mapping data. Based on the density distribution mapping data, the spatial positional relationship between the initial feature points is analyzed to obtain the correlation degree. The topological relationship of the sub-regions with the correlation degree lower than the preset correlation degree threshold is adjusted to obtain the reorganized spatial structure data. Feature points with higher integrity than a preset integrity standard in the recombined spatial structure data are retained and their geometric and distribution attributes are extracted to obtain a structured feature dataset.

[0011] In one optional implementation, the step of extracting the geometric topological relationships of feature points from the structured feature dataset and performing protective encoding to generate an encrypted feature combination includes: Based on the distribution characteristics of the structured feature dataset, the geometric topological relationship data of each feature point is obtained. The distance and angle between the feature points are compared with the pre-established rule base to obtain the weight allocation result. The weight allocation result is subjected to multi-level data transformation to obtain the topological constraints between the feature points at each transformation level, and the priority order of encryption encoding is determined. The original feature points are grouped according to the priority order, and the stable features of the geometric topological relationship are extracted to obtain the encrypted feature combination. The encrypted feature combination is verified a second time to obtain the feature combination data.

[0012] In one optional implementation, the encrypted feature sets for which the visibility of the feature point density distribution map is lower than a preset visibility threshold are adjusted to obtain a fine-tuned feature set, including: Based on the density distribution characteristics of the feature points, the density distribution map of the feature points is scanned to obtain visibility assessment data; For visibility assessment data that are lower than a preset confidence threshold, adjust the matching error tolerance range to obtain adjusted distribution characteristic data; The adjusted distribution characteristic data is subjected to secondary optimization processing to obtain dynamic change data of matching error value and tolerance interval, and to determine the new positional relationship of feature points in the distribution map. Based on the new positional relationship, group and integrate the data, extract stable data that conforms to the distribution characteristics, and generate the fine-tuned feature set.

[0013] In one optional implementation, comparing the refined feature set with a pre-established palm vein template library to generate a palm vein authentication result includes: Obtain the feature vector data of the target user from the fine-tuned feature set; The feature vector data is compared one by one with the target user's historical feature vectors extracted from a pre-established palm vein template library to obtain a similarity score; The similarity scores are sorted, and the similarity scores of the top-ranked scores are compared with the preset confidence threshold of the recognition results to determine whether they meet the threshold requirements. If the authentication is successful, an authentication pass flag is generated; if the authentication fails, an authentication failure flag is generated, and the flag data is stored in a temporary database. Based on the flag data in the temporary database, the final verification status data is obtained, classified, and generated into a palm vein authentication result, which is then output to a designated storage unit.

[0014] Secondly, the present invention provides a privacy encryption system based on palm vein features, comprising: The image correction module is used to acquire the original palm vein image data, perform light compensation and reflectivity adjustment on the original palm vein image, and obtain an initial corrected image. The saturation analysis module is used to perform layered processing on the initial corrected image, determine the saturation fluctuation benchmark value of each layer, divide the initial corrected image into blocks according to the saturation fluctuation benchmark value, mark the area where the saturation fluctuation exceeds the preset fluctuation threshold as the saturation uneven area, and generate a saturation distribution map. The dynamic correction module is used to mark areas of uneven saturation and, in conjunction with the local brightness contrast coefficient, adjust the parameters of the areas of uneven saturation to obtain a balanced corrected image. The feature extraction module extracts key palm vein feature points from the equalization and correction image, constructs a feature point density distribution map, and performs spatial structure reconstruction on the feature point density distribution map to obtain a structured feature dataset. The data encryption module is used to extract the geometric topological relationship of feature points from the structured feature dataset, perform multi-layer data transformation and protective encoding on the geometric topological relationship of feature points, and generate encrypted feature combinations. The optimization module is used to determine whether the visibility of the feature point density distribution map is lower than a preset visibility threshold. If it is lower, the matching error tolerance range is adjusted, and the encrypted feature combination is optimized a second time to obtain a fine-tuned feature set. The verification module is used to compare the fine-tuned feature set with a pre-established palm vein template library to generate a palm vein authentication result.

[0015] Thirdly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the computer-readable storage medium method described in any one of the above.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention performs systematic preprocessing on the original palm vein image data. By equalizing the grayscale values ​​of the ambient light interference factor and skin reflectivity correction value, it ensures that all original image data has been cleaned, standardized and verified, thereby guaranteeing the integrity, consistency and accuracy of the image data. This process effectively eliminates image quality problems caused by unstable acquisition environment, and lays a reliable foundation for subsequent feature extraction and encryption processing; (2) This invention achieves classification and standardization based on dynamic saturation analysis and multi-layer data transformation mechanism. Through saturation distribution mapping and dynamic grayscale stretching technology, palm vein features are classified, labeled, and structurally reorganized according to regional saturation differences, generating standardized encrypted feature combinations. This invention maps palm vein feature points hierarchically to different encrypted spaces according to geometric topological relationships, enabling feature data to be stored in a template library by category, achieving efficient retrieval and management.

[0017] (3) Based on these standardized and tokenized feature data, this invention can further support confidence assessment for identity verification, dynamic detection of anti-counterfeiting attacks, and iterative optimization of encryption templates. The labels of each feature not only strengthen the privacy protection mechanism, but also provide a clear structured basis for subsequent dynamic queries and comparisons. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a privacy encryption method based on palm vein features provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a privacy encryption system module based on palm vein features provided in an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] In the field of biometrics, privacy protection technology is directly related to the security of personal identity information and the construction of a social trust system. With the acceleration of digitalization, how to achieve privacy protection while ensuring recognition accuracy has become a core challenge.

[0021] Most existing technologies rely on a single encryption mode or fixed process, failing to adapt dynamically to the characteristics of biometric data, resulting in limited protection capabilities in complex environments or under diverse attacks. Furthermore, existing technologies neglect the synergy between the quality of feature data visualization and protection mechanisms during data acquisition and processing, leading to inadequate privacy protection.

[0022] To address the aforementioned issues, the following specific embodiments will provide a detailed description and explanation of a privacy encryption method based on palm vein features provided in this application.

[0023] Reference Figure 1 The first embodiment of the present invention provides a privacy encryption method based on palm vein features, comprising the following steps: S101, acquire the original palm vein image and perform preprocessing to obtain the initial corrected image; S102, the initial corrected image is segmented and detected, regions with uneven saturation are marked, and a saturation distribution map is generated; S103, Based on the saturation distribution map and the local brightness contrast coefficient, the parameters of the uneven saturation region are adjusted to generate a balanced correction image; S104, extract key palm vein feature points from the balanced correction image, construct a feature point density distribution map and perform spatial structure reconstruction to obtain a structured feature dataset; S105, extract the geometric topological relationship of feature points from the structured feature dataset and perform protective encoding to generate encrypted feature combinations; S106, if the visibility of the feature point density distribution map is lower than the preset visibility threshold of the encrypted feature combination, adjust the matching error tolerance range to obtain a fine-tuned feature set; S107, compare the fine-tuned feature set with the pre-established palm vein template library to generate palm vein authentication results.

[0024] The light influence factor is the global average brightness of the image, used to collect the average brightness value of the image. In practical applications, a brightness threshold range of 50 to 200 can be set. The initial value of the skin reflectivity in the first correction data is adjusted to 0.6, while the corresponding value in the preset correction coefficient table is 0.5, with a deviation of 0.1. To address the problem of insufficient skin texture in the image, the gray-level difference in the texture area is increased from 5 to 15 by enhancing local contrast. This process effectively improves the visibility of the image and lays the foundation for subsequent processing.

[0025] In step S101, the original palm vein image is acquired and preprocessed to obtain an initial corrected image, including: S1011, acquire the original palm vein image, and when the light influence factor exceeds the preset influence range, perform light compensation on the original palm vein image to obtain the first correction data; S1012, The reflectivity of the first correction data is adjusted using a preset correction coefficient table to determine the second correction data; S1013, adjust the grayscale distribution and enhance the contrast of the second correction data to obtain the initial correction image.

[0026] In step S1011, the original palm vein image is acquired. When the light influence factor exceeds the preset influence range, the original palm vein image is light compensated to obtain the first correction data. It should be noted that the light influence factor refers to the phenomenon of uneven image brightness caused by factors such as changes in ambient light during image acquisition. Light compensation is a correction method used during palm vein image acquisition to address image brightness issues caused by ambient light interference by adjusting the brightness value to bring the image within a preset reasonable range.

[0027] In one implementation, a high-resolution camera captures a skin image during the acquisition of raw image data. Ambient light interference can cause uneven image brightness; for example, under indoor lighting, some areas of the image may be too bright, while shadow areas may be too dark. A brightness threshold range of 50 to 200 can be set to address ambient light interference and the light influence factor. Specific values ​​are acquired using an ambient light sensor under different lighting conditions, and the difference between these values ​​is calculated to determine the light influence factor. If the average brightness value of the acquired image is significantly lower than the lower threshold, light compensation processing is required. The first correction data is obtained by increasing the brightness value to 80.

[0028] Effective calibration is crucial for improving the quality of palm vein images, enhancing the accuracy of feature extraction, and optimizing privacy encryption, ensuring the efficient and reliable operation of the entire biometric recognition system. This process effectively improves image visibility, laying the foundation for subsequent processing.

[0029] In step S1012, the reflectivity of the first correction data is adjusted using a preset correction coefficient table to determine the second correction data.

[0030] It should be noted that reflectance is a quantitative indicator that measures the skin's ability to reflect light, and is used to correct uneven illumination in palm vein images. The preset correction coefficient table is a reference data table for image processing. By providing standardized correction parameters, it helps adjust the image's grayscale distribution, reflectance, and other characteristics, thereby optimizing image quality, improving processing efficiency, and enhancing the accuracy of feature extraction.

[0031] In one implementation, the skin reflectivity in the first correction data is adjusted. For example, the initial skin reflectivity value is 0.6, while the corresponding value in the preset correction coefficient table is 0.5, with a deviation of 0.1. If the deviation exceeds the specified range of 0.05, a second correction is required through an iterative adjustment algorithm, decreasing the value by 0.02 each time. After the iterative adjustment algorithm, the reflectivity is finally determined to be 0.52, resulting in the second correction data.

[0032] It can effectively correct the unevenness of skin reflectivity and reduce image brightness differences caused by reflectivity deviations. This adjustment process can also optimize the basic data quality of privacy encryption mechanisms, ensuring the efficient and reliable operation of the palm vein image processing and recognition system under different lighting conditions.

[0033] In step S1013, the grayscale distribution of the second correction data is adjusted and the contrast is enhanced to obtain an initial correction image.

[0034] It should be noted that grayscale distribution adjustment is an image processing technique that adjusts the distribution of grayscale values ​​in an image. Adjusting the grayscale distribution involves changing the grayscale values ​​of pixels in the image, achieved through histogram equalization or grayscale transformation methods. By statistically analyzing and mapping grayscale values, the brightness and contrast of the image are made more uniform.

[0035] In one implementation, during the grayscale distribution adjustment stage, based on the second correction data, it is found that the image grayscale values ​​are unevenly distributed, with some areas having grayscale values ​​concentrated in the low-value range. Histogram equalization is used to redistribute the grayscale values, expanding their coverage to a wider range, for example, from 0 to 100 to 0 to 180, resulting in the third correction data. To address the issue of insufficiently clear skin texture in the image, image contrast enhancement is used. By enhancing local contrast, the grayscale difference in textured areas is increased from 5 to 15, ultimately yielding the initial corrected image.

[0036] Through the aforementioned multi-step correction, the impact of environmental factors on images can be effectively reduced, ensuring that doctors can make judgments based on more accurate image data. This multi-level correction improves the reliability of the analysis and provides important support for related fields.

[0037] In step S102, the initial corrected image is segmented and detected to mark regions with uneven saturation and generate a saturation distribution map, including: S1021, perform layered processing on the initial corrected image, obtain the saturation fluctuation threshold range corresponding to the data fluctuation of each layer, and determine the saturation fluctuation reference value of each layer. S1022, using the saturation fluctuation reference value and the region segmentation grid density, block processing is performed to obtain multiple regions; S1023, mark the regions in the multiple regions whose saturation fluctuation exceeds a preset fluctuation threshold as the saturation unevenness regions, and obtain the marked region set; S1024, Scan the marked region set to obtain saturation distribution data and obtain saturation distribution characteristics; S1025, The saturation distribution features are integrated to obtain a saturation distribution mapping map.

[0038] In step S1021, the initial corrected image is processed in layers to obtain the saturation fluctuation threshold range corresponding to the data fluctuation of each layer, and the saturation fluctuation reference value of each layer is determined.

[0039] It should be noted that the initial correction image layering process divides the image into multiple gray levels by processing the pixel gray level distribution range of the initial correction image, and then analyzes the data fluctuations. For saturation fluctuations, a reasonable threshold range is set as a judgment criterion.

[0040] In one implementation, when processing the pixel grayscale distribution range of the initial calibrated image, the image is divided into multiple grayscale levels. If the image's grayscale values ​​range from 0 to 255, the tool divides it into 5 intervals, each covering approximately 51 grayscale values. For each interval, the data fluctuation is analyzed, and a reasonable threshold range, such as 5 to 15, is set as a judgment criterion.

[0041] By employing layered grayscale interval division and dynamic grid marking, regions with uneven saturation are accurately identified and located, significantly improving the overall consistency of the image and the reliability of subsequent analysis. This layered processing facilitates a more detailed observation of the characteristics of different grayscale regions in the image, providing a precise basis for subsequent processing. Regions with uneven saturation are then marked into a set for targeted processing.

[0042] In step S1022, the saturation fluctuation reference value is combined with the region segmentation grid density to perform block processing, resulting in multiple regions.

[0043] It should be noted that the saturation fluctuation baseline value is the average saturation value within each grayscale range, used to quantify regional consistency. Combining the data distribution within each grayscale range, a representative value is extracted as a baseline. The image is then divided into blocks using a grid, thereby adjusting the image resolution. Specifically, this block-based processing divides the image into local regions using grid density as the unit, detecting and marking saturation or feature point anomalies block by block, achieving a precise localization and optimization image processing method.

[0044] In one implementation, the baseline value for saturation fluctuation can be determined by combining the data distribution within each grayscale range and extracting a representative value as the baseline. For example, if the saturation fluctuation values ​​are mostly concentrated between 8 and 12 within a certain grayscale range, then the median value can be taken as the baseline. This is one method for determining the baseline value for saturation fluctuation. It's worth noting that during block processing, a grid density needs to be set, and the unit size needs to be determined based on the image resolution and processing accuracy requirements to form a regular grid array, generating grid coordinates so that the image is divided into multiple small blocks.

[0045] Block processing combined with saturation fluctuation baseline values ​​allows for detailed local analysis and adjustment of the image. This transforms the "whole image" into "blocks." Dynamically adjusting the saturation of these regions enhances the overall consistency and visual quality of the image, thereby improving image quality and providing a more reliable and accurate data foundation for subsequent feature extraction and analysis.

[0046] In step S1023, regions in the multiple regions whose saturation fluctuation exceeds a preset fluctuation threshold are marked as saturation uneven regions, thus obtaining a set of marked regions.

[0047] It's important to note that saturation fluctuation exceeding a preset fluctuation threshold refers to a situation where, during image processing, the degree of saturation change in a certain region or block exceeds a pre-defined allowable range. Saturation is a crucial attribute of image color, describing the purity or intensity of a color. Ideally, saturation should be uniform. A region set refers to a collection of image regions with similar characteristics or attributes identified and marked using specific algorithms or tools during image processing. These regions can be defined based on different criteria.

[0048] In one implementation, the entire image is divided into non-overlapping grid cells according to the multiple regions, arranged from left to right and top to bottom. Each cell serves as a unique coordinate index, resulting in marked anomalous blocks. The saturation fluctuation and feature point density within each grid are compared to a preset threshold; units exceeding the threshold are marked as "uneven saturation regions" or "sparse feature regions," forming a "marker set" that can be directly used for subsequent local correction. Regions with saturation fluctuation values ​​exceeding the upper threshold are marked as uneven saturation regions, forming a marker set. Based on this benchmark, a grid partitioning tool is used to divide the image into blocks. The grid density can be adjusted according to the image resolution, for example, one grid cell is 10x10 pixels. If the saturation fluctuation value of a certain region reaches 18, significantly exceeding the upper threshold of 15, it is marked as an uneven saturation region, forming a marker set for subsequent targeted processing.

[0049] This processing method helps improve image quality, ensuring that the uniformity of image saturation is crucial for the accuracy of feature extraction and recognition. By identifying and labeling these regions, operations can be performed more precisely on specific parts of the image, thereby improving processing efficiency and effectiveness.

[0050] In step S1024, the marked region set is scanned to obtain saturation distribution data and obtain saturation distribution characteristics.

[0051] It's important to note that saturation distribution data is essentially a numerical map of the saturation levels in different parts of the image. This data is used to quickly pinpoint areas of abnormal illumination / reflection, guiding subsequent corrections. Accurately exposing areas with illumination defects allows for targeted corrections, significantly improving the uniformity and recognition accuracy of vein images. The marked region aggregation scanning process utilizes a block-based scanning frequency adjustment. Increasing the scanning frequency for each area with uneven saturation allows for a more comprehensive understanding of the problem area, resulting in a more detailed saturation distribution map and a more complete understanding of the specific situation within the problem area.

[0052] In one implementation, when performing high-frequency scanning on the marked region set, the scanning frequency for each region with uneven saturation can be increased by segmenting the scan, for example, from once per second to five times per second, to obtain more detailed saturation distribution data. The scan results might show that saturation values ​​within a certain region range from 5 to 20, exhibiting extremely uneven distribution. This meticulous scanning method helps to gain a more comprehensive understanding of the specific situation in the problem area, providing data support for further adjustments.

[0053] This method mainly obtains detailed saturation distribution data through high-frequency block scanning, accurately locates areas of abnormal illumination and reflection, provides data support for subsequent correction, and significantly improves image uniformity and recognition accuracy.

[0054] In step S1025, the saturation distribution features are integrated to obtain a saturation distribution mapping map.

[0055] It should be noted that integrating and processing saturation distribution features can quickly locate problem areas. By mapping the saturation data of each region onto a unified distribution map, and adjusting the mapping logic, a saturation distribution mapping map can be generated.

[0056] In one implementation, when integrating saturation distribution characteristics, image mapping can be used to generate a unified distribution map mapping the saturation data of each region. For example, if the saturation values ​​in one region are concentrated in a high-value area, while in another region they are relatively low, after adjustment through mapping logic, an intuitive saturation distribution map can be generated, clearly showing which parts of the image exhibit unevenness. Generating this map helps to quickly locate problem areas, providing an intuitive reference for subsequent optimization and ensuring a more balanced saturation distribution in each region.

[0057] This meticulous processing method can effectively improve the overall consistency of the image, laying a solid foundation for subsequent skin feature analysis.

[0058] In step S103, based on the saturation distribution map and the local brightness contrast coefficient, parameters are adjusted for the uneven saturation region to generate a balanced correction image, including: S1031, Extract the specific location of the uneven saturation region in the saturation distribution map; S1032, the specific location of the uneven area is initially marked, and the marked area that exceeds the preset deviation threshold is determined as the target adjustment area; S1033, Based on the location data of the target adjustment area, obtain the corresponding local brightness distribution information and perform quantization processing, divide the target adjustment area into multiple target adjustment sub-regions, and continuously integrate each target adjustment sub-region to obtain the comparison coefficient value. S1034, adjust the corresponding correction parameters for the contrast coefficient value of the target adjustment sub-region, increase the grayscale stretching amplitude of the target adjustment sub-region with a contrast coefficient lower than the preset contrast threshold, and determine the adjusted parameter configuration; S1035, Based on the adjusted parameter configuration and combined with the overall saturation distribution and brightness distribution data, the equalization correction image is generated.

[0059] In step S1031, the specific locations of the uneven saturation regions in the saturation distribution map are extracted.

[0060] It should be noted that the specific locations of the uneven saturation regions are the coordinates of pixel regions whose saturation significantly deviates from the threshold, marked by a saturation distribution map. This coordinates are used to accurately locate the illumination / reflection defects that need correction, ensuring the targeted effectiveness of subsequent grayscale stretching and equalization processing. The judgment of saturation deviation relies not only on numerical values ​​but also on the overall visual characteristics of the image to ensure the accuracy of the marking. This enhances the security of vein feature extraction and encryption.

[0061] In one implementation, when analyzing the saturation distribution map, image analysis tools are used to extract the specific location data of uneven regions. For example, in an image, the saturation value of the upper left corner is significantly higher than other areas, reaching 85, while the average value of the surrounding areas is only 50, exceeding a preset threshold of 20. By comparing this value to the preset threshold, this region can be initially identified as uneven, and further analysis can be performed to determine whether it meets the adjustment criteria.

[0062] By analyzing the saturation distribution map and identifying uneven regions, image quality can be optimized, and the accuracy of feature extraction and recognition system can be significantly improved, which is crucial for improving the performance of the entire biometric recognition system.

[0063] In step S1032, the specific locations of the uneven regions are initially marked, and the marked regions that exceed the preset deviation threshold are determined as target adjustment regions.

[0064] It should be noted that the specific location of the uneven region is the set of coordinates of the grid blocks marked as uneven saturation in the saturation distribution map. This method can allocate processing resources more rationally, helps to accurately locate the problem area, and provides data support for dynamic processing.

[0065] In one implementation, adjusting the marked regions that exceed a preset deviation threshold can be achieved through a multi-dimensional evaluation mechanism. This mechanism considers both the region's area proportion and saturation deviation. Regions with small saturation deviations but large area proportions that significantly impact the overall image quality are prioritized for adjustment; otherwise, they are temporarily excluded from the target adjustment list.

[0066] For example, a marked region may have a saturation deviation of 25, but its area only accounts for 2% of the total image area, which may have a minor impact on the overall visual appearance, so it will not be included in the target adjustment region for the time being. On the other hand, another region has a deviation of 22 and its area accounts for 15%, so it will be prioritized as the target adjustment region. This method can allocate processing resources more rationally.

[0067] This operation can accurately locate the areas in the image that need adjustment, ensuring that subsequent correction operations such as grayscale stretching are more targeted and effective. It provides a more accurate and higher-quality image foundation for subsequent feature extraction and privacy encryption, thereby improving the reliability and accuracy of the entire biometric recognition system.

[0068] In step S1033, based on the location data of the target adjustment area, the corresponding local brightness distribution information is obtained and quantized. The target adjustment area is divided into multiple target adjustment sub-regions, and the comparison coefficient value is obtained by continuously integrating each target adjustment sub-region.

[0069] It should be noted that the target adjustment area is used to obtain local brightness distribution information, which is then quantified using a brightness contrast coefficient calculation tool. Quantization transforms local brightness differences into measurable contrast coefficients, allowing for rapid threshold-based judgment and guidance for subsequent corrections. This converts subjective visual differences into objective values, accurately identifying brightness defect areas and ensuring targeted dynamic grayscale stretching.

[0070] In this invention, local brightness distribution information of the target adjustment area is obtained and quantified using a brightness contrast coefficient calculation tool. Assuming the average brightness of a sub-region within the target area is 30, while the average brightness of adjacent regions is 60, the calculated contrast coefficient is 0.5, which is lower than the preset threshold of 0.8, indicating that the brightness of this sub-region is too low and requires further adjustment.

[0071] This quantification process helps to accurately locate problem areas and provides data support for dynamic processing.

[0072] In step S1034, the corresponding correction parameters are adjusted for the contrast coefficient value of the target adjustment sub-region, the grayscale stretching amplitude of the target adjustment sub-region with a contrast coefficient lower than the preset contrast threshold is increased, and the adjusted parameter configuration is determined.

[0073] It should be noted that the adjustment range needs to be fine-tuned according to the overall characteristics of the image to avoid overstretching and distortion. This method can effectively improve the visual clarity of local areas. A sub-region refers to the smallest processing unit obtained by further subdividing the target adjustment area according to a preset grid, which is used to independently calculate the average brightness and determine the grayscale stretching separately.

[0074] For example, for dynamic processing of grayscale stretching, the correction parameters can be adjusted based on the contrast coefficient value. Suppose the contrast coefficient of a certain sub-region is 0.6, which is lower than the threshold of 0.8, then the grayscale stretching amplitude can be increased from the default 1.2 to 1.5 to enhance the contrast effect of the area.

[0075] By dynamically adjusting the grayscale stretching amplitude, the contrast effect of light and dark areas in sub-regions with contrast coefficients below a preset threshold can be significantly enhanced, thereby improving the visual clarity of local areas while avoiding distortion caused by overstretching. This process not only optimizes image quality but also provides a more reliable foundation for subsequent feature extraction and privacy encryption.

[0076] In step S1035, the equalization correction image is generated based on the adjusted parameter configuration and combined with the overall saturation distribution and brightness distribution data.

[0077] It should be noted that parameter configuration involves setting adjustable values ​​such as gain, offset, and threshold for grayscale stretching, used to precisely control the enhancement magnitude and range of low-contrast sub-regions. By quantizing and controlling the enhancement magnitude, the brightness of low-contrast sub-regions is accurately matched to the global equalization, avoiding overexposure or distortion. The equalization-corrected image is a high-quality image with clear details, uniform distribution, and a more natural visual effect after grayscale, saturation, and brightness optimization, providing a reliable foundation for subsequent analysis and applications.

[0078] In one implementation, the process of generating a balanced corrected image mainly includes acquiring an initial image, adjusting grayscale distribution, saturation distribution, and brightness distribution, and finally applying the adjusted parameter configurations to generate the image. This is achieved through steps such as histogram equalization, dynamic saturation adjustment, and brightness contrast coefficient optimization. Specifically, when applying the adjusted parameter configurations, image generation tools can be used to apply the correction parameters to the target adjustment area, while simultaneously combining overall saturation and brightness distribution data to ensure that the final balanced corrected image is visually more harmonious. For example, the saturation deviation in the target area decreases from 25 to 10, and the brightness contrast coefficient increases from 0.5 to 0.85, significantly improving the uniformity of the overall image. This processing method enhances the overall image quality, laying the foundation for subsequent applications.

[0079] By applying the adjusted parameter configuration and combining it with overall saturation and brightness distribution data to generate a balanced and corrected image, it is possible to accurately match the brightness of low-contrast sub-regions with global balance, avoid overexposure or distortion, thereby significantly improving the uniformity and overall quality of the image and providing a solid foundation for subsequent applications.

[0080] In step S104, key palm vein feature points are extracted from the equalized and corrected image, a feature point density distribution map is constructed, and spatial structure is reorganized to obtain a structured feature dataset, including: S1041, Obtain the original point data from the equalization correction image, perform edge detection and noise filtering to obtain an initial feature point set; S1042, Construct a density distribution map based on the initial feature point set, divide the density distribution map into multiple density sub-regions and calculate the distribution density of each sub-region to determine the density distribution mapping data; S1043, Based on the density distribution mapping data, analyze the spatial positional relationship between the initial feature points to obtain the correlation degree, and adjust the topological relationship of the sub-regions with the correlation degree lower than the preset correlation degree threshold to obtain the reorganized spatial structure data; S1044, the feature points in the recombined spatial structure data with integrity higher than the preset integrity standard are retained and their geometric and distribution attributes are extracted to obtain a structured feature dataset.

[0081] In step S1041, the original point data is obtained from the equalization correction image, and edge detection and noise filtering are performed to obtain an initial feature point set.

[0082] It should be noted that obtaining the original point data refers to extracting key feature point information of the palm veins from the equalized and corrected image. This data represents the distribution location of the palm veins. Edge detection and noise filtering are then performed using image processing tools to remove interfering points, resulting in an initial set containing the core features.

[0083] In one implementation, when processing the raw point data of key palm veins in a balanced and corrected image, basic point information can be extracted from the image first. For example, with an image resolution of 800x600 pixels, the raw point data may contain thousands of coordinate points representing the distribution locations of palm veins. In the initial processing stage, edge detection tools can be used to identify the boundaries of the vein patterns, while noise filtering removes interfering points caused by light or skin surface impurities, ultimately resulting in an initial set containing core features, such as approximately 500 key points. The purpose of this process is to ensure the data quality for subsequent analysis.

[0084] This process lays the foundation for subsequent feature extraction and analysis, ensuring the accuracy and reliability of feature points, thereby improving the performance and accuracy of the entire biometric identification system.

[0085] In step S1042, a density distribution map is constructed based on the initial feature point set, and the density distribution map is divided into multiple density sub-regions and the distribution density of each sub-region is calculated to determine the density distribution mapping data.

[0086] It should be noted that constructing a density distribution map from the initial feature point set involves organizing and analyzing the key palm vein feature points extracted from the balanced and corrected image. By calculating the number of feature points in each sub-region and visualizing them, a chart showing the distribution density of feature points is created. Distribution density refers to the distribution of the number of palm vein feature points within a specific region, reflecting the degree of concentration of feature points in the image.

[0087] In one implementation, when constructing a density distribution map for an initial set of feature points, the image is divided into multiple sub-regions. Within each sub-region, the number of feature points is counted, and the distribution density of that region is calculated. The density distribution map integrates the distribution density data of each sub-region into a single image, visually displaying the distribution of feature points. The image can be divided into a 10x10 grid, with each small grid covering a certain area, such as 80x60 pixels. Density distribution mapping data is obtained by calculating the number of feature points within each grid. For example, a grid might contain 20 points, while adjacent grids might only have 5 points.

[0088] This density difference reflects the uneven distribution of veins, providing a basis for subsequent spatial relationship analysis. Constructing a density distribution map helps to visually understand the concentrated and sparse areas of feature points.

[0089] In step S1043, the spatial relationship between the initial feature points is analyzed based on the density distribution mapping data to obtain the correlation degree. The topological relationship of the sub-regions with the correlation degree lower than the preset correlation degree threshold is adjusted to obtain the reorganized spatial structure data.

[0090] It should be noted that density distribution mapping data is data that calculates the number and distribution of feature points in each sub-region of an image, used to visually demonstrate the density of feature points in different regions. The spatial relationship between feature points is their geometric attribute in the image. Correlation measures the degree of correlation or connection between feature points in space. Sub-region topology adjustment involves adjusting and optimizing the connections between feature points within a specific sub-region based on their distribution and correlation.

[0091] In one implementation, when analyzing the spatial relationships between feature points, the coordinates of the feature points are first obtained through feature point detection, then the geometric attributes between the feature points are calculated to evaluate their correlation. The topological relationships within sub-regions are then adjusted as needed. Geometric calculations are used to assess the correlation between points. For example, if the preset correlation threshold is 0.7, and the average correlation between points within a certain sub-region is only 0.4, then topological adjustments are required. These adjustments are made by connecting adjacent points or re-dividing the boundaries of sub-regions to create a more compact spatial structure.

[0092] This adjustment optimizes the organization of feature points, making subsequent data integration smoother.

[0093] S1044, the feature points in the recombined spatial structure data with integrity higher than the preset integrity standard are retained and their geometric and distribution attributes are extracted to obtain a structured feature dataset.

[0094] It should be noted that feature points with preset integrity criteria refer to the set of feature points that meet specific conditions after feature point extraction and spatial relationship analysis. These conditions typically include the number of feature points, distribution density, geometric attributes, and correlation with other feature points. Feature points with preset integrity criteria are high-quality feature point sets that have been screened and optimized for subsequent feature extraction, privacy encryption, and authentication processing.

[0095] In one implementation, when converting the reconstructed spatial structure data into a structured feature dataset, the geometric attributes of each feature point can be extracted, such as the distance and angle between points, as well as distribution attributes, such as the density ratio of points in the grid. Assuming the integrity criterion requires each sub-region to contain at least three valid points, if a region does not meet this requirement, it may be necessary to supplement with virtual points or merge data from adjacent regions. The resulting structured feature dataset will contain complete geometric and distribution information, laying the foundation for subsequent image analysis.

[0096] This multi-layered processing approach improves data accuracy while enriching the adaptability of the solution through optional implementation methods. Ultimately, these processing techniques collectively support the complete transformation process from raw images to structured data, providing reliable support for subsequent applications.

[0097] In step S105, the geometric topological relationships of feature points are extracted from the structured feature dataset and protected by encoding to generate encrypted feature combinations, including: S1051, Based on the distribution characteristics of the structured feature dataset, obtain the geometric topological relationship data of each feature point, and compare the distance and angle between the feature points with the help of a pre-established rule base to obtain the weight allocation result; S1052, Perform multi-layer data transformation on the weight allocation result, obtain the topological constraint conditions between the feature points at each layer of transformation, and determine the priority order of encryption encoding; S1053, group the original feature points according to the priority order, extract the stable features of the geometric topological relationship, and obtain the encrypted feature combination; S1054, perform a second verification on the encrypted feature combination to obtain feature combination data.

[0098] In step S1051, based on the distribution characteristics of the structured feature dataset, the geometric topological relationship data of each feature point is obtained, and the distance and angle between the feature points are compared with the help of a pre-established rule base to obtain the weight allocation result.

[0099] It's important to note that a pre-established rule base refers to a collection of predefined rules used to guide the extraction, analysis, and processing of feature points. The construction of a rule base typically involves abstracting and formalizing domain-specific knowledge to ensure that the rules accurately reflect the logic and requirements of practical applications. A structured feature dataset organizes the attributes of palm vein feature points into a clearly defined tabular format, which can be directly used for subsequent comparison and encryption. Geometric topological relationship data typically reflects the relative positions and connections between feature points. Geometric topological relationship data digitizes the relative positions and connections between feature points, giving the palm vein structure measurable uniqueness.

[0100] In one implementation, when processing the distribution characteristics of a structured feature dataset, one can begin by analyzing the geometric topological relationships of the feature points to understand their spatial distribution patterns. Geometric topological relationship data typically reflects the relative positions and connections between feature points, providing a foundation for subsequent comparison and analysis. For example, in a palm vein feature analysis scenario, the dataset contains 100 feature points, and the coordinates of each feature point and its connections to neighboring points are recorded.

[0101] By analyzing this data, we can make a preliminary judgment on which feature points are more concentrated in space and which are more dispersed, thus laying the foundation for weight allocation.

[0102] In one possible implementation, the comparison of distances and angles between feature points can be performed using a pre-established comparison rule base. This rule base may contain weight standards corresponding to different distance and angle ranges; for example, feature point pairs with a distance of 2 to 5 units might have a weight of 0.8, while those with angles between 30 and 60 degrees might have a weight of 0.6. Through such rule-based comparison, preliminary weight allocation results can be quickly obtained.

[0103] The advantage of this method is that it can quickly filter out feature point pairs with high correlation, saving resources for subsequent processing.

[0104] In step S1052, the weight allocation result is subjected to multi-layer data transformation to obtain the topological constraints between the feature points at each layer of transformation, and the priority order of encryption encoding is determined.

[0105] It should be noted that the weight allocation result refers to the weight distribution table formed by reassigning new weight values ​​to each feature point after weight calculation based on the geometric topological relationships of the feature points. This table directly determines which points will be prioritized, retained, or merged in subsequent encryption, comparison, or reorganization processes, serving as the primary or secondary quantitative basis. Topological constraints are the "geometric rules" that must be forcibly satisfied in each step of data transformation or reorganization.

[0106] In one implementation, distance and angle comparisons between feature points can be performed using a pre-established comparison rule base. The rule base may contain weight standards corresponding to different distance and angle ranges; feature point pairs with a distance of 2 to 5 units have a weight of 0.8, while those with an angle between 30 and 60 degrees have a weight of 0.6. Preliminary weight allocation results can be obtained quickly. The advantage of this method is that it can quickly filter out feature point pairs with high correlation, saving resources for subsequent processing. In the multi-layer data transformation process, feature point data can be mapped to different transformation spaces. In each layer of transformation, topological constraints are obtained, such as ensuring that the relative distance between feature points does not change significantly after transformation. These constraints help determine the priority order of encryption encoding. Assuming that some feature points have an encoding priority of 0.9 or higher, meeting a preset threshold of 0.85, these feature points can be preferentially encoded for protection. This layered processing method effectively improves data security.

[0107] In step S1054, the encrypted feature combination is verified a second time to obtain feature combination data.

[0108] It should be noted that feature point grouping refers to dividing feature points into several subsets according to weights or topological constraints, so that they can be processed independently in groups during encryption or comparison, improving computational efficiency and ensuring structural stability within each group. Encrypted feature combination is a method of encrypting palm vein feature points after weighted grouping and topological encoding for secure comparison. It ensures the uniqueness of vein topology while completely hiding the original image information, achieving both high recognition rate and strong anti-leakage capability.

[0109] In one possible implementation, a protective coding mechanism can group feature points for processing, enhancing data security and privacy through feature point extraction, multi-layer data transformation, and a series of encryption techniques. It encodes the geometric topological relationships of feature points to generate encrypted feature combinations. For example, 100 feature points can be divided into 5 groups of 20 points each, and stable geometric features can be extracted from each group, such as a group where the average distance between feature points is consistently 3.2 units. These stable features can serve as the basis for generating encrypted feature combinations. This grouping process ensures the preservation of local stability of the feature data, thereby improving the overall reliability of the data. Secondary verification of the encrypted feature combinations can be performed using general data processing tools. If, during the verification process, a feature combination with a geometric deviation of less than 0.1 units is found to meet the requirements, it is considered to be compliant.

[0110] This verification method further ensures the accuracy of the feature combination data, providing a guarantee for the final generation of results that meet the requirements for protecting geometric topological relationships. While protecting the data, it also provides high-quality foundational data for subsequent applications.

[0111] In step S106, if the visibility of the feature point density distribution map is lower than a preset visibility threshold for the encrypted feature combination, the matching error tolerance interval is adjusted to obtain a fine-tuned feature set, including: S1061, Based on the density distribution characteristics of the feature points, scan the density distribution map of the feature points to obtain visibility assessment data; S1062, Adjust the matching error tolerance range for visibility assessment data that are lower than the preset confidence threshold to obtain adjusted distribution characteristic data; S1063, perform secondary optimization processing on the adjusted distribution characteristic data to obtain dynamic change data of matching error value and tolerance interval, and determine the new positional relationship of feature points in the distribution map; S1064, group and integrate the data according to the new positional relationship, extract stable data that conforms to the distribution characteristics, and generate the fine-tuned feature set.

[0112] In step S1061, geometric topological relationship data of each feature point is obtained based on the distribution characteristics of the structured feature dataset.

[0113] It's important to note that the feature point density distribution map is a "sparse / dense map of vein feature points" displayed in the form of a grid or heatmap. The color of each grid represents the number of palm vein feature points in that area, allowing you to immediately see the concentrated and blank areas of the vein patterns. The density distribution map makes the "density" of veins clear at a glance, quickly filtering out false features, identifying core areas, improving recognition speed, and reducing interference.

[0114] In one implementation approach, when dealing with tasks related to feature point density distribution characteristics, one can start by understanding the spatial distribution patterns of the data to comprehend its manifestation in a specific scenario. Feature point density distribution maps typically reflect the degree of clustering and sparsity of data points in space; this distribution characteristic is particularly important in scenarios such as palm vein feature analysis. Suppose in a specific application, the dataset contains 200 feature points, and the distribution map shows that some areas have high density while other areas are sparse. A preliminary scan of the distribution map using general data processing tools can obtain visibility assessment data; for example, the visibility assessment value for a certain area is 0.75, while the preset threshold is 0.8. Since this does not meet the standard, further adjustments are needed.

[0115] This adjustment helps optimize the distribution of feature points, improve the accuracy and reliability of feature extraction, and thus enhance the performance and accuracy of the entire biometric identification system.

[0116] In step S1062, the visibility assessment data that is lower than the preset confidence threshold is adjusted to obtain the matching error tolerance range, thereby obtaining the adjusted distribution characteristic data.

[0117] It's important to note that the visibility assessment data quantifies the readability of the density distribution map into a score of 0-1. Scores below the threshold trigger secondary optimization to ensure reliable recognition. Visibility assessment data identifies "blurred areas" early, driving error correction and directly improving the pass rate while reducing the false rejection rate. The preset confidence threshold is a minimum acceptable score manually set by the system before the recognition process begins. It determines whether the visibility of the feature point density distribution map is sufficiently reliable; scores below this value trigger secondary optimization or reject the match. The confidence threshold acts like a "passing grade," automatically filtering out low-quality images, ensuring accurate recognition while avoiding false positives.

[0118] In one implementation, when visibility assessment data falls below a threshold, error correction can be performed by adjusting the tolerance range of the matching error value. Assuming the initial tolerance range is 0.1 to 0.3 units, it is adjusted to 0.05 to 0.35 units. The distribution map is then reanalyzed using a general error calculation tool to obtain the corrected data. This adjustment helps improve the overall consistency of the distribution map, especially in cases of uneven density distribution, enabling a better capture of the true distribution patterns of feature points.

[0119] In step S1063, the adjusted distribution characteristic data is subjected to secondary optimization processing to obtain dynamic change data of matching error value and tolerance interval, and to determine the new positional relationship of feature points in the distribution map.

[0120] It should be noted that the dynamic changes in the matching error value and tolerance range are dynamic numerical pairs between the "error magnitude" calculated in real time during the feature point matching process and the "error range" allowed by the system, which are used to automatically determine when to stop adjusting and lock the final feature position. By comparing the error value and tolerance range in real time, the system can quickly converge to the optimal match without sacrificing accuracy, significantly improving the recognition speed.

[0121] In one implementation, when performing secondary optimization on the adjusted distribution characteristic data, the visibility of the feature point density distribution is evaluated to determine whether it meets the preset confidence threshold for the recognition result. If the visibility is below the threshold, the distribution characteristics of the feature points are recalculated by adjusting the matching error tolerance interval, and the positions of the feature points are fine-tuned. Assuming that during the optimization process, the matching error value decreases while the tolerance interval remains within 0.05 to 0.35 units, it indicates that the error has been effectively controlled. Through this analysis of dynamically changing data, the new positional relationships of feature points in the distribution map can be determined; for example, some feature points may move from the edge region to the center region, forming a more compact distribution pattern.

[0122] Secondary optimization of the adjusted distribution characteristic data can improve the uniformity and visibility of feature point distribution, reduce misjudgments, and thus improve the overall performance and recognition accuracy of the biometric recognition system.

[0123] S1064, group and integrate the data according to the new positional relationship, extract stable data that conforms to the distribution characteristics, and generate the fine-tuned feature set.

[0124] It is important to know that the distribution characteristics are statistical descriptions of the spatial clustering, sparsity, uniformity, and local density differences of palm vein feature points in an image. These characteristics are used to determine whether the overall structure is clear and stable, and thus decide whether optimization or secondary correction is needed.

[0125] One implementation involves grouping and integrating new positional relationships. Feature points are divided into several groups using general data integration. For example, 200 feature points are divided into 4 groups of 50 each, and stable data conforming to distribution characteristics are extracted, such as a group where the average density of feature points is consistently above 0.9 units. This stable data serves as the basis for generating the final fine-tuned set, ensuring that local characteristics of the data are preserved while improving the reliability of the overall distribution.

[0126] In step S107, the fine-tuned feature set is compared with a pre-established palm vein template library to generate a palm vein authentication result, including: S1071, Obtain the feature vector data of the target user from the fine-tuned feature set; S1072, The feature vector data is compared one by one with the target user's historical feature vectors extracted from the pre-established palm vein template library to obtain a similarity score; S1073, Sort the similarity scores according to the similarity scores, compare the similarity scores of the top-ranked scores with the preset confidence threshold of the recognition results, and determine whether they meet the threshold requirements. S1074, if the conditions are met, an authentication pass flag is generated; if the conditions are not met, an authentication failure flag is generated, and the flag data is stored in a temporary database. S1075, based on the flag data in the temporary database, obtain the final verification status data, perform classification processing, generate the palm vein authentication result, and output it to the designated storage unit.

[0127] In step S1071, the feature vector data of the target user is obtained from the fine-tuned feature set.

[0128] It should be noted that the fine-tuned feature set refers to a high-quality set of feature points that, after secondary optimization, have a more uniform distribution, higher visibility, and meet the preset confidence threshold for recognition results. This is achieved through steps such as adjusting the matching error tolerance range and optimizing the feature point positions. Feature vector data refers to the digital representation of key feature points extracted from palm vein images. These feature points, after processing and optimization, are converted into a set of multi-dimensional values ​​used to describe the geometric and topological characteristics of the palm veins. Feature vector data is the core data used for identity verification in biometric recognition systems.

[0129] In one implementation, within the business domain of palm vein recognition, the process of acquiring and comparing feature vector data can be understood in principle as achieving identity verification through the digital representation of biometric features. Palm veins, as a unique biometric feature, can have their texture information converted into feature vectors using infrared scanning equipment. Specifically, in the feature vector extraction stage, assuming the target user's palm vein image is scanned, the system generates a feature vector containing multi-dimensional data, such as the location and angle of vein branch points. For historical feature vectors in the template library, the system stores feature vector data from multiple past scans, such as five historical records, each containing similar multi-dimensional data. Using a vector distance calculation tool, the system calculates the distance difference between the current feature vector and these five historical feature vectors, obtaining a similarity score. The score ranges from 0 to 1, with higher values ​​indicating greater similarity.

[0130] This process improves the accuracy and reliability of identification, supports dynamic adjustment and optimization, enhances system security and privacy protection, and provides detailed verification feedback. These effects collectively improve the overall performance and user experience of the biometric identification system.

[0131] In step S1072, the feature vector data is compared one by one with the target user's historical feature vectors extracted from the pre-established palm vein template library to obtain a similarity score.

[0132] It should be noted that the similarity score ranking and threshold comparison process refers to the system sorting the similarity scores of the current feature vector and historical feature vectors from high to low, and comparing the top-ranked scores with the preset confidence threshold of the recognition result to determine whether they meet the identity verification requirements.

[0133] In one implementation, during the similarity score ranking and threshold comparison stage, the system arranges the five scores from highest to lowest, assuming the results are 0.92, 0.88, 0.85, 0.80, and 0.75. If the preset confidence threshold for the recognition result is 0.90, then the top-ranked score of 0.92 meets the requirements, and the system will generate a preliminary identity verification pass flag.

[0134] This approach ensures the reliability of the verification results, avoids interference from low-similarity data on the final result, and improves the performance and security of the biometric identification system.

[0135] In step S1073, the similarity scores are sorted according to the similarity scores, and the similarity scores of the top-ranked scores are compared with the preset confidence threshold of the recognition results to determine whether they meet the threshold requirements.

[0136] It should be noted that the similarity score is a list of 0-1 scores obtained by comparing the current palm vein feature vector with each historical vector in the template library one by one. The higher the score, the more similar the two are, and it is used to directly determine whether the identity verification is successful. The preset confidence threshold for the recognition result is a "passing grade" score set in advance by the system. Only when the highest similarity score obtained from the comparison is greater than or equal to this value is the identity verification considered successful; otherwise, it is rejected or a secondary verification is triggered. This threshold acts like a "safety gate," automatically filtering low similarity matches, ensuring high accuracy while preventing false recognition.

[0137] In one implementation, during the similarity score ranking and threshold comparison stage, the system arranges the five scores from highest to lowest, assuming the results are 0.92, 0.88, 0.85, 0.80, and 0.75. If the preset confidence threshold for the recognition result is 0.90, then the top-ranked score of 0.92 meets the requirement, and the system generates a preliminary identity verification pass flag. This method ensures the reliability of the verification results and avoids interference from low-similarity data on the final result.

[0138] In step S1074, if the authentication is successful, an authentication success flag is generated; if the authentication is unsuccessful, an authentication failure flag is generated, and the flag data is stored in a temporary database.

[0139] It should be noted that generating an authentication success flag means that in a biometric identification system, when the similarity score between the current feature vector and historical feature vectors meets a preset confidence threshold for the identification result, the system generates a flag indicating that the user's authentication has been successful. By comparing similarity scores and judging the threshold, the system confirms the legitimacy of the user's identity, records the verification result, supports subsequent system operations, and ensures the security and reliability of the system.

[0140] In one implementation, when processing verification flags and subsequent processes, if the score does not reach a threshold (e.g., the highest score is 0.88, but below 0.90), the system generates an authentication failure flag and stores it in a temporary database. For the flags in the temporary database, a logical judgment tool further categorizes and processes them, for example, marking the failure flag as requiring secondary verification to ensure the integrity of the verification process.

[0141] This verifies the legitimacy of the user's identity, records the verification results, supports subsequent system operations, and ensures the security and reliability of the system.

[0142] In step S1075, the final verification status data is obtained based on the flag data in the temporary database, and the data is classified and processed to generate the palm vein authentication result, which is then output to the designated storage unit.

[0143] It should be noted that classification processing refers to further processing and classifying the verification flags based on the authentication results, so that the system can take appropriate follow-up actions according to different verification results.

[0144] In one implementation, during the final output stage, the system generates the final palm vein authentication result based on the categorized verification status data and outputs it to a designated storage unit, such as a cloud database or local storage device. This output method facilitates subsequent querying and data management, while also providing users with clear verification feedback. From an extended solution perspective, if the initial verification fails, the system can also combine other auxiliary data, such as auxiliary identity information entered by the user, with the palm vein data for joint verification.

[0145] This approach further enhances the accuracy and security of verification, which is particularly important in scenarios with high security requirements. Overall, the close integration and multi-layered processing of these steps ensure the efficiency and reliability of palm vein recognition services, providing a solid technical guarantee for user authentication.

[0146] In summary, this invention discloses a privacy encryption method based on palm vein features, comprising: The original palm vein image is acquired and preprocessed to obtain the initial corrected image; The initial corrected image is segmented and detected to mark regions with uneven saturation and generate a saturation distribution map. Based on the saturation distribution map and the local brightness contrast coefficient, the parameters of the uneven saturation region are adjusted to generate a balanced correction image. Key palm vein feature points are extracted from the balanced and corrected image, a feature point density distribution map is constructed, and spatial structure is reorganized to obtain a structured feature dataset. The geometric topological relationships of feature points are extracted from the structured feature dataset and protected by encoding to generate encrypted feature combinations; If the visibility of the feature point density distribution map is lower than the preset visibility threshold for the encrypted feature combination, the matching error tolerance interval is adjusted to obtain a fine-tuned feature set; The fine-tuned feature set is compared with a pre-established palm vein template library to generate a palm vein authentication result.

[0147] Reference Figure 2 The second embodiment of the present invention provides a privacy encryption system based on palm vein features, comprising: The image correction module is used to acquire the original palm vein image data, perform light compensation and reflectivity adjustment on the original palm vein image, and obtain an initial corrected image. The saturation analysis module is used to perform layered processing on the initial corrected image, determine the saturation fluctuation benchmark value of each layer, divide the initial corrected image into blocks according to the saturation fluctuation benchmark value, mark the area where the saturation fluctuation exceeds the preset fluctuation threshold as the saturation uneven area, and generate a saturation distribution map. The dynamic correction module is used to mark areas of uneven saturation and, in conjunction with the local brightness contrast coefficient, adjust the parameters of the areas of uneven saturation to obtain a balanced corrected image. The feature extraction module extracts key palm vein feature points from the equalization and correction image, constructs a feature point density distribution map, and performs spatial structure reconstruction on the feature point density distribution map to obtain a structured feature dataset. The data encryption module is used to extract the geometric topological relationship of feature points from the structured feature dataset, perform multi-layer data transformation and protective encoding on the geometric topological relationship of feature points, and generate encrypted feature combinations. The optimization module is used to determine whether the visibility of the feature point density distribution map is lower than a preset visibility threshold. If it is lower, the matching error tolerance range is adjusted, and the encrypted feature combination is optimized a second time to obtain a fine-tuned feature set. The verification module is used to compare the fine-tuned feature set with a pre-established palm vein template library to generate a palm vein authentication result.

[0148] It should be noted that the privacy encryption system based on palm vein features provided in this embodiment of the invention is used to execute all the process steps of the privacy encryption method based on palm vein features in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0149] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a privacy encryption system based on palm vein features. When the processor executes the computer program, it implements the steps in the aforementioned privacy encryption method embodiments based on palm vein features, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as an image correction module.

[0150] For example, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0151] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0152] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0153] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0154] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0155] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0156] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that 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 for those skilled in the art.

Claims

1. A privacy encryption method based on palm vein features, characterized in that, Executed by a computer, including: The original palm vein image is acquired and preprocessed to obtain the initial corrected image; The initial corrected image is segmented and detected to mark regions with uneven saturation and generate a saturation distribution map. Based on the saturation distribution map and the local brightness contrast coefficient, the parameters of the uneven saturation region are adjusted to generate a balanced correction image. Key palm vein feature points are extracted from the balanced and corrected image, a feature point density distribution map is constructed, and spatial structure is reorganized to obtain a structured feature dataset. The geometric topological relationships of feature points are extracted from the structured feature dataset and protected by encoding to generate encrypted feature combinations; If the visibility of the feature point density distribution map is lower than the preset visibility threshold for the encrypted feature combination, the matching error tolerance interval is adjusted to obtain a fine-tuned feature set; The fine-tuned feature set is compared with a pre-established palm vein template library to generate a palm vein authentication result.

2. The privacy encryption method based on palm vein features according to claim 1, characterized in that, The process of acquiring raw palm vein image data and performing preprocessing to obtain an initial corrected image includes: Acquire the original palm vein image, and when the light influence factor exceeds the preset influence range, perform light compensation on the original palm vein image to obtain the first correction data; The reflectance of the first correction data is adjusted using a preset correction coefficient table to determine the second correction data; The grayscale distribution of the second correction data is adjusted and the contrast is enhanced to obtain the initial correction image.

3. The privacy encryption method based on palm vein features according to claim 1, characterized in that, The process of segmenting and detecting regions in the initial corrected image, marking regions with uneven saturation, and generating a saturation distribution map includes: The initial corrected image is processed in layers to obtain the saturation fluctuation threshold range corresponding to the data fluctuation of each layer, and the saturation fluctuation reference value of each layer is determined. By combining the saturation fluctuation benchmark value with the region segmentation grid density, block processing is performed to obtain multiple regions; The regions in the multiple regions whose saturation fluctuations exceed a preset fluctuation threshold are marked as the saturation unevenness regions, thus obtaining a set of marked regions; The marked region set is scanned to obtain saturation distribution data and thus saturation distribution characteristics. The saturation distribution features are integrated to obtain a saturation distribution mapping map.

4. The privacy encryption method based on palm vein features according to claim 1, characterized in that, The step of adjusting parameters in the uneven saturation region based on the saturation distribution map and local brightness contrast coefficient to generate a balanced corrected image includes: Extract the specific locations of the uneven saturation regions in the saturation distribution map; The specific locations of the uneven areas are initially marked, and the marked areas that exceed the preset deviation threshold are determined as the target adjustment areas; Based on the location data of the target adjustment area, the corresponding local brightness distribution information is obtained and quantified. The target adjustment area is divided into multiple target adjustment sub-regions, and the contrast coefficient value is obtained by continuously integrating each target adjustment sub-region. Adjust the corresponding correction parameters for the contrast coefficient value of the target adjustment sub-region, increase the grayscale stretching amplitude of the target adjustment sub-region with a contrast coefficient lower than the preset contrast threshold, and determine the adjusted parameter configuration; Based on the adjusted parameter configuration, and combined with the overall saturation and brightness distribution data, the equalization correction image is generated.

5. The privacy encryption method based on palm vein features according to claim 1, characterized in that, The process involves extracting key palm vein feature points from the balanced and corrected image, constructing a feature point density distribution map, and reconstructing its spatial structure to obtain a structured feature dataset, including: The original point data is obtained from the equalization and correction image, and edge detection and noise filtering are performed to obtain an initial set of feature points; A density distribution map is constructed based on the initial set of feature points, and the density distribution map is divided into multiple density sub-regions. The distribution density of each sub-region is calculated to determine the density distribution mapping data. Based on the density distribution mapping data, the spatial positional relationship between the initial feature points is analyzed to obtain the correlation degree. The topological relationship of the sub-regions with the correlation degree lower than the preset correlation degree threshold is adjusted to obtain the reorganized spatial structure data. Feature points with higher integrity than a preset integrity standard in the recombined spatial structure data are retained and their geometric and distribution attributes are extracted to obtain a structured feature dataset.

6. The privacy encryption method based on palm vein features according to claim 1, characterized in that, The step of extracting the geometric topological relationships of feature points from the structured feature dataset and performing protective encoding to generate encrypted feature combinations includes: Based on the distribution characteristics of the structured feature dataset, the geometric topological relationship data of each feature point is obtained. The distance and angle between the feature points are compared with the pre-established rule base to obtain the weight allocation result. The weight allocation result is subjected to multi-level data transformation to obtain the topological constraints between the feature points at each transformation level, and the priority order of encryption encoding is determined. The original feature points are grouped according to the priority order, and the stable features of the geometric topological relationship are extracted to obtain the encrypted feature combination. The encrypted feature combination is verified a second time to obtain the feature combination data.

7. The privacy encryption method based on palm vein features according to claim 1, characterized in that, The encrypted feature sets for which the visibility of the feature point density distribution map is lower than a preset visibility threshold are adjusted to improve the matching error tolerance range, resulting in a fine-tuned feature set, including: Based on the density distribution characteristics of the feature points, the density distribution map of the feature points is scanned to obtain visibility assessment data; For visibility assessment data that are lower than a preset confidence threshold, adjust the matching error tolerance range to obtain adjusted distribution characteristic data; The adjusted distribution characteristic data is subjected to secondary optimization processing to obtain dynamic change data of matching error value and tolerance interval, and to determine the new positional relationship of feature points in the distribution map. Based on the new positional relationship, group and integrate the data, extract stable data that conforms to the distribution characteristics, and generate the fine-tuned feature set.

8. The privacy encryption method based on palm vein features according to claim 1, characterized in that, The step of comparing the fine-tuned feature set with a pre-established palm vein template library to generate a palm vein authentication result includes: Obtain the feature vector data of the target user from the fine-tuned feature set; The feature vector data is compared one by one with the target user's historical feature vectors extracted from a pre-established palm vein template library to obtain a similarity score; The similarity scores are sorted, and the similarity scores of the top-ranked scores are compared with the preset confidence threshold of the recognition results to determine whether they meet the threshold requirements. If the authentication is successful, an authentication pass flag is generated; if the authentication fails, an authentication failure flag is generated, and the flag data is stored in a temporary database. Based on the flag data in the temporary database, the final verification status data is obtained, classified, and generated into a palm vein authentication result, which is then output to a designated storage unit.

9. A privacy encryption system based on palm vein features, characterized in that, include: The image correction module is used to acquire the original palm vein image data, perform light compensation and reflectivity adjustment on the original palm vein image, and obtain an initial corrected image. The saturation analysis module is used to perform layered processing on the initial corrected image, determine the saturation fluctuation benchmark value of each layer, divide the initial corrected image into blocks according to the saturation fluctuation benchmark value, mark the area where the saturation fluctuation exceeds the preset fluctuation threshold as the saturation uneven area, and generate a saturation distribution map. The dynamic correction module is used to mark areas of uneven saturation and, in conjunction with the local brightness contrast coefficient, adjust the parameters of the areas of uneven saturation to obtain a balanced corrected image. The feature extraction module extracts key palm vein feature points from the equalization and correction image, constructs a feature point density distribution map, and performs spatial structure reconstruction on the feature point density distribution map to obtain a structured feature dataset. The data encryption module is used to extract the geometric topological relationship of feature points from the structured feature dataset, perform multi-layer data transformation and protective encoding on the geometric topological relationship of feature points, and generate encrypted feature combinations. The optimization module is used to determine whether the visibility of the feature point density distribution map is lower than a preset visibility threshold. If it is lower, the matching error tolerance range is adjusted, and the encrypted feature combination is optimized a second time to obtain a fine-tuned feature set. The verification module is used to compare the fine-tuned feature set with a pre-established palm vein template library to generate a palm vein authentication result.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the privacy encryption method based on palm vein features as described in any one of claims 1 to 8.

Images