Small-area fingerprint identification method, device and medium

Abstract

The application relates to the fingerprint identification technical field and discloses a small-area fingerprint identification method, equipment and medium. The method comprises the following steps: acquiring a to-be-identified fingerprint image; performing ridge feature point extraction on the to-be-identified fingerprint image to obtain a feature point set; the feature point set comprises first feature points and / or second feature points, the first feature points are bifurcation points and / or end points of the fingerprint ridge line, and the second feature points are maximum curvature points of the fingerprint ridge line; performing image feature extraction on the to-be-identified fingerprint image based on the feature points in the feature point set to obtain a plurality of corresponding local fingerprint image features; performing feature matching on the local fingerprint image features and preset target fingerprint image features, and generating a fingerprint identification result of the to-be-identified fingerprint image based on the feature matching result. The application can improve the identification accuracy of small-area fingerprint identification.

Description

TECHNICAL FIELD

[0001] The present application relates to the fingerprint identification technical field, and especially to a small-area fingerprint identification method, device and medium. BACKGROUND

[0002] Although the fingerprint identification technology in the related art is mature and reliable, the required sensor area is large and the cost is high, and there is no competitive advantage on mobile devices or embedded devices. Under the action of market competition, the existing fingerprint identification technology begins to turn to small-area fingerprint identification. However, the amount of traditional fingerprint feature information in small-area fingerprint is reduced, resulting in a decrease in recognition accuracy. How to improve the accuracy of small-area fingerprint identification has become a difficult problem to be solved. The current small-area fingerprint identification method is mainly based on machine vision, such as sift, surf, ORB, Akaze, etc. Although the number of feature points constructed by this kind of method is large, it solves the problem of the small number of traditional fingerprint feature points, but its stability is poor and the recognition performance will be sharply reduced due to factors such as season, molting, dry and wet. SUMMARY

[0003] The purpose of the present application is to provide a small-area fingerprint identification method, device and medium, aiming to improve the recognition accuracy of small-area fingerprint identification.

[0004] The embodiment of the present application provides a small-area fingerprint identification method, comprising:

[0005] obtaining a to-be-identified fingerprint image;

[0006] extracting ridge line feature points from the to-be-identified fingerprint image to obtain a feature point set; the feature point set includes first feature points and / or second feature points, the first feature points are bifurcation points and / or end points of fingerprint ridges, and the second feature points are maximum curvature points of fingerprint ridges;

[0007] extracting image features from the to-be-identified fingerprint image based on the feature points in the feature point set to obtain a plurality of corresponding local fingerprint image features;

[0008] performing feature matching on the local fingerprint image features and preset target fingerprint image features, and generating a fingerprint identification result of the to-be-identified fingerprint image based on the feature matching result.

[0009] In some embodiments, the extracting ridge line feature points from the to-be-identified fingerprint image comprises:

[0010] searching for first feature points in the to-be-identified fingerprint image;

[0011] judging whether the number of the first feature points exceeds a feature point number threshold;

[0012] If not, search for a second feature point in the to-be-identified fingerprint image, and form the feature point set with the first feature point and the second feature point.

[0013] If not, search for a second feature point in the to-be-identified fingerprint image, and form the feature point set with the first feature point and the second feature point.

[0014] In some embodiments, the searching for the first feature point in the to-be-identified fingerprint image comprises:

[0015] Performing fingerprint ridge feature extraction on the to-be-identified fingerprint image to obtain a ridge feature map;

[0016] Calculating a total color difference of neighborhood pixels of a target pixel point; the target pixel point is a pixel point constituting a fingerprint ridge in the ridge feature map, and the total color difference of neighborhood pixels is a sum of pixel color differences between each two adjacent neighborhood pixel points of the target pixel point;

[0017] Based on the total color difference of neighborhood pixels, determining a bifurcation point and an end point of the fingerprint ridge in the ridge feature map to obtain the first feature point.

[0018] In some embodiments, the searching for the second feature point in the to-be-identified fingerprint image comprises:

[0019] Performing direction feature extraction on the to-be-identified fingerprint image to obtain a direction feature map;

[0020] Performing curvature feature extraction on the direction feature map to obtain a curvature feature map;

[0021] Performing Gaussian smoothing on the curvature feature map, searching for a pixel point with a maximum pixel value after Gaussian smoothing, and obtaining the second feature point.

[0022] In some embodiments, the performing image feature extraction on the to-be-identified fingerprint image based on the feature points in the feature point set comprises:

[0023] Taking a feature point in the feature point set as a center, performing screenshot on the to-be-identified fingerprint image to obtain a plurality of original feature point maps;

[0024] Performing image feature extraction on the original feature point maps to obtain the local fingerprint image features.

[0025] In some embodiments, the taking a feature point in the feature point set as a center, performing screenshot on the to-be-identified fingerprint image comprises:

[0026] perform rigid transformation on the to-be-identified fingerprint image based on position information and direction information of the feature points in the feature point set, to obtain a transformed feature map with the feature points in the feature point set as centers;

[0027] perform intercepting and invalid region labeling on the transformed feature map, to obtain the original feature point map.

[0028] In some embodiments, the feature matching of the local fingerprint image feature with the preset target fingerprint image feature comprises:

[0029] calculating a feature matching degree between the local fingerprint image feature and the target fingerprint image feature;

[0030] based on the feature matching degree, matching the local fingerprint image feature generated based on the first feature point with the target fingerprint image feature generated based on the first feature point, and / or matching the local fingerprint image feature generated based on the second feature point with the target fingerprint image feature generated based on the second feature point, and generating a fingerprint recognition result of the to-be-identified fingerprint image based on a matching result.

[0031] In some embodiments, the calculation of the feature matching degree between the local fingerprint image feature and the target fingerprint image feature comprises:

[0032] based on a rigid transformation mapping relationship between the local fingerprint image feature and the target fingerprint image feature, verifying consistency of feature point position dimensions and feature point direction dimensions of both the local fingerprint image feature and the target fingerprint image feature, to obtain a consistency verification result;

[0033] when the consistency verification result meets a preset consistency condition, calculating the feature matching degree based on an average similarity, a highest similarity and a lowest similarity between the local fingerprint image feature and the target fingerprint image feature.

[0034] Embodiments of the present application also provide an electronic device, which comprises a memory and a processor, the memory stores a computer program, and the processor implements the small-area fingerprint identification method described above when executing the computer program.

[0035] Embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the small-area fingerprint identification method described above.

[0036] The beneficial effects of this application are as follows: By constructing a feature point set by fusing stable ridge topological features with highly discriminative curvature features, and implementing localized feature processing and matching strategies, the accuracy decline problem caused by insufficient traditional feature information in small-area fingerprint recognition is effectively addressed. The diversified design of the feature point set enhances information representation within a limited area, the local feature extraction strategy reduces the negative impact of environmental factors on feature stability, and the feature matching process strengthens the tolerance to deformation and noise through targeted comparison of local features. Thus, high-reliability fingerprint recognition performance is maintained under the condition of limited sensor area, avoiding the performance fluctuation problem caused by existing machine vision methods relying on pixel-level changes, and improving the recognition accuracy of small-area fingerprint recognition. Attached Figure Description

[0037] Figure 1 This diagram illustrates the application environment of the small-area fingerprint recognition method provided in the embodiments of this application.

[0038] Figure 2 This is a flowchart of a small-area fingerprint recognition method provided in an embodiment of this application.

[0039] Figure 3 This is a flowchart of a method for extracting ridge feature points from a fingerprint image to be identified, provided in an embodiment of this application.

[0040] Figure 4 This is a flowchart of a method for extracting image features from a fingerprint image to be identified, provided in an embodiment of this application.

[0041] Figure 5 This is a flowchart of a method for feature matching of local fingerprint image features with preset target fingerprint image features, provided in an embodiment of this application.

[0042] Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0044] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and drawings are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0045] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0046] The small-area fingerprint recognition method provided in this application can be executed by a computer device, which can be a terminal device or a server. The terminal device includes, but is not limited to, mobile phones, computers, smart home appliances, vehicle terminals, and aircraft. The server can be a standalone physical server, a server cluster consisting of multiple physical servers, a distributed system, or a cloud server. Furthermore, the information, data, and signals involved in this application's embodiments are all authorized by the relevant parties or have been fully authorized by all parties, and the collection, use, and processing of related data comply with the relevant laws, regulations, and standards of the relevant countries and regions.

[0047] While existing fingerprint recognition technologies are mature and reliable, they require large sensor areas, resulting in high costs and a lack of competitive advantage in mobile or embedded devices. Driven by market competition, existing fingerprint recognition technologies are shifting towards smaller fingerprint recognition areas. However, the reduced amount of traditional fingerprint feature information in smaller fingerprint areas leads to decreased recognition accuracy. Improving the accuracy of small-area fingerprint recognition has become a pressing problem. Current small-area fingerprint recognition methods are primarily based on machine vision, such as SIFT, SURF, ORB, and Akaze. While these methods generate a large number of feature points, solving the problem of limited feature points in traditional fingerprints, their stability is poor, and recognition performance can drop sharply due to factors such as season, peeling skin, and wetness / dryness.

[0048] In traditional fingerprint recognition technologies, the limited sensor area significantly reduces the density of traditional ridge topological features in small fingerprint images, resulting in insufficient feature points to support reliable identification. Simultaneously, the pixel-level variations relied upon by machine vision-based feature extraction methods are susceptible to skin condition interference. When a user's finger is seasonally dry, peeling, or damp, the geometric properties of feature points shift uncontrollably, leading to increased false matching rates and frequent rejections, thus affecting the overall system reliability and environmental adaptability. For example, in fingerprint unlocking scenarios on mobile devices, fluctuations in skin dryness or localized peeling due to environmental humidity or daily wear and tear result in insufficient ridge bifurcation and endpoint extraction using traditional methods. Furthermore, feature points generated by SIFT or ORB algorithms drift due to skin texture distortion, causing the feature matching process to fail to accurately associate with the target fingerprint template. This leads to repeated prompts for re-entry, interrupting the unlocking operation and directly compromising the continuity of the device's security verification process and the smoothness of the user experience.

[0049] If the above problems are not solved, the robustness of the identification system in complex environments will continue to weaken, the uncertainty of feature matching will lead to the risk of failure of the security verification mechanism, thereby limiting the large-scale application of small-area fingerprint sensors in resource-constrained devices and hindering the functional integrity and market competitiveness of biometric technology in embedded scenarios.

[0050] Based on this, embodiments of this application provide a small-area fingerprint recognition method, device, and medium. By introducing the maximum curvature point, bifurcation point, and / or endpoint of the fingerprint ridge as supplementary feature points, and combining a feature point quality evaluation mechanism to dynamically optimize feature point selection, the robustness of feature extraction and matching accuracy of small-area fingerprint images are effectively improved, thereby enhancing the recognition accuracy of small-area fingerprint recognition.

[0051] Figure 1 This diagram illustrates the application environment of the small-area fingerprint recognition method provided in this embodiment. (See attached diagram.) Figure 1 This method is applied to a small-area fingerprint recognition system. The system includes a terminal 110 and a server 120. The terminal 110 and server 120 are connected via a network. The terminal 110 can be a desktop terminal or a mobile terminal; the mobile terminal can be at least one of a mobile phone, tablet, or laptop. The server 120 can be a standalone server or a server cluster consisting of multiple servers. The terminal 110 sends a fingerprint image to be recognized to the server 120. The server 120 acquires the fingerprint image, extracts ridge feature points from the fingerprint image to obtain a feature point set, extracts image features from the fingerprint image based on the feature points in the feature point set, obtains several corresponding local fingerprint image features, performs feature matching between the local fingerprint image features and preset target fingerprint image features, and generates a fingerprint recognition result based on the feature matching result. The feature point set includes a first feature point and / or a second feature point, where the first feature point is the bifurcation point and / or endpoint of the fingerprint ridge, and the second feature point is the point of maximum curvature of the fingerprint ridge.

[0052] It should be understood that Figure 1 The application scenarios shown are merely examples. In practical applications, the small-area fingerprint recognition method provided in this application embodiment can also be applied to other scenarios. For example, the above-described small-area fingerprint recognition method can be directly applied to terminal 110. Terminal 110 is used to acquire a fingerprint image to be recognized, extract ridge feature points from the fingerprint image to be recognized to obtain a feature point set, extract image features from the fingerprint image to be recognized based on the feature points in the feature point set to obtain several corresponding local fingerprint image features, perform feature matching between the local fingerprint image features and the preset target fingerprint image features, and generate a fingerprint recognition result of the fingerprint image to be recognized based on the feature matching result.

[0053] See Figure 2 In one embodiment, a small-area fingerprint recognition method is provided. The execution subject of the method can be either a terminal or a server, including but not limited to steps S201 to S204.

[0054] Step S201: Obtain the fingerprint image to be identified.

[0055] Step S202: Extract ridge feature points from the fingerprint image to be identified to obtain a feature point set.

[0056] The feature point set includes a first feature point and / or a second feature point, wherein the first feature point is the bifurcation point and / or endpoint of the fingerprint ridge, and the second feature point is the point of maximum curvature of the fingerprint ridge.

[0057] Step S203: Based on the feature points in the feature point set, perform image feature extraction on the fingerprint image to be identified to obtain several corresponding local fingerprint image features.

[0058] Step S204: Perform feature matching between the local fingerprint image features and the preset target fingerprint image features, and generate the fingerprint recognition result of the fingerprint image to be recognized based on the feature matching result.

[0059] The first feature point refers to the key location of topological changes in the fingerprint ridge structure. It can be achieved by binarizing and skeletonizing the fingerprint image and then identifying the bifurcation and endpoints. For example, the ridge can be converted into a single-pixel-width skeleton by thinning algorithm and the tri-points or isolated endpoints can be detected, or morphological operations can be used to locate the bifurcation region and termination point in the binary image.

[0060] The second feature point refers to the location where the curvature of the fingerprint ridge changes most significantly. It can be achieved by parametrically fitting the ridge skeleton and calculating the curvature extrema. For example, the maximum value of the curvature derivative can be found after fitting the ridge path based on spline curves, or the curvature can be estimated and the peak position can be screened by using the rate of change of the tangent vector of the discrete point set.

[0061] Local fingerprint image features refer to the microstructural information of a fingerprint represented by a local region centered on a feature point. This can be achieved by cropping a fixed-size window and applying local descriptors. For example, a 16×16 pixel region can be cropped around the feature point, and then the histogram of oriented gradients can be calculated. Alternatively, the cropped region can be normalized to grayscale, and local binary pattern features can be extracted. The process of matching local fingerprint image features with preset target fingerprint image features and generating fingerprint recognition results based on the matching results can be achieved by calculating the Euclidean distance between feature vectors. For example, the k-nearest neighbor algorithm can be used to compare the similarity of local features, or a support vector machine model can be used to classify and determine the matching degree.

[0062] The fingerprint image to be identified is acquired as the basic data source for the identification process. This image is collected by a small optical sensor, such as the fingerprint module integrated into the power button of a mobile device. Subsequently, ridge feature point extraction is applied to the fingerprint image to generate a feature point set, which includes a first feature point and / or a second feature point. The extraction of the first feature point relies on the inherent topological characteristics of the fingerprint ridges, exhibiting high intrinsic stability and being less susceptible to external factors such as seasonal changes, skin peeling, or environmental humidity. The extraction of the second feature point is based on the geometric properties of the ridge curvature changes, providing highly discriminative supplementary information within a small area, thereby significantly enhancing information density under conditions where the number of feature points is limited. Further, based on each feature point in the feature point set, image feature extraction is performed to generate several corresponding local fingerprint image features. This process crops a local area centered on the feature point and performs feature representation, ensuring that the features focus on the microscopic morphology of the fingerprint ridges and avoiding redundancy and noise interference introduced by global feature extraction in small-area scenes.

[0063] This application constructs a feature point set by integrating traditional ridge features with newly added curvature features, and implements localized feature processing and matching to effectively address the challenges of limited information and insufficient feature stability in small-area fingerprint recognition. Specifically, the extraction of the first feature point relies on the inherent topological stability of the ridge, making it less susceptible to interference from surface dryness, wetness, or peeling. The introduction of the second feature point stems from the inherent geometric characteristics of ridge curvature changes, providing highly discriminative supplementary information within a small area. Local feature extraction based on feature points focuses on the microscopic morphology of the fingerprint ridge, avoiding redundancy and noise interference from global feature extraction in small-area scenarios. The feature matching process, through targeted comparison of local features and combined with the inherent correlation between the position and orientation of feature points, enhances tolerance to deformation and noise, thereby maintaining high-reliability recognition under conditions of limited information.

[0064] As a specific embodiment, in a smartphone fingerprint recognition scenario, the fingerprint image to be recognized is captured by a miniature optical sensor. During ridge feature point extraction, the first feature point is identified as a bifurcation point and endpoint using an image thinning algorithm, and the second feature point is determined as the point of maximum curvature through curvature calculation. In the image feature extraction stage, a 16×16 pixel local region is cropped centered on each feature point, and a local binary mode algorithm is applied to generate local fingerprint image features. Subsequently, the local fingerprint image features are matched with preset target fingerprint image features, the feature matching degree is calculated, and the fingerprint recognition result is determined based on the feature matching degree.

[0065] Therefore, this method constructs a feature point set by fusing stable ridge topological features with highly discriminative curvature features, and implements localized feature processing and matching strategies. This effectively addresses the accuracy decline problem caused by insufficient traditional feature information in small-area fingerprint recognition. Specifically, the diversified design of the feature point set enhances information representation within a limited area, the local feature extraction strategy reduces the negative impact of environmental factors on feature stability, and the feature matching process strengthens tolerance to deformation and noise through targeted comparison of local features. Thus, it maintains high-reliability fingerprint recognition performance under sensor area constraints, avoiding the performance fluctuation problem caused by existing machine vision methods that rely on pixel-level changes.

[0066] In some embodiments of this application, ridge feature point extraction is proposed to obtain a feature point set. However, during its implementation, when the number of first feature points in a small fingerprint image is insufficient, the information content of the feature point set is too low, which cannot support the reliability of the subsequent matching process, resulting in a significant decrease in fingerprint recognition accuracy. Therefore, this application further proposes a method for ridge feature point extraction from the fingerprint image to be recognized.

[0067] See Figure 3 In one embodiment, the method for extracting ridge feature points from the fingerprint image to be identified includes, but is not limited to, steps S301 to S304.

[0068] Step S301: Search for the first feature point in the fingerprint image to be identified.

[0069] Step S302: Determine whether the number of the first feature points exceeds the feature point number threshold.

[0070] If the limit is exceeded, proceed to step S303; if the limit is not exceeded, proceed to step S304.

[0071] Step S303: Perform feature point quality evaluation on the first feature point, and select the first feature points that meet the feature point quality evaluation conditions to form a feature point set.

[0072] Step S304: Search for a second feature point in the fingerprint image to be identified, and form a feature point set with the first feature point and the second feature point.

[0073] The feature point quantity threshold is a preset value used to quantify the sufficiency of feature points. It can be achieved by dynamically calculating based on image resolution or by statistical analysis of historical data. The purpose is to accurately identify sparse feature point scenes and avoid the adaptive defects caused by blindly relying on a fixed number.

[0074] Feature point quality evaluation refers to the process of assessing the reliability of feature points. It can be achieved by using methods based on pixel contrast gradient analysis or neighborhood structure consistency verification. The purpose is to eliminate inferior points caused by noise interference or blurred areas, and ensure the high reliability of the feature point set.

[0075] Searching for second feature points refers to the strategy of supplementing ridge morphology sensitive points when there are insufficient feature points. It can be achieved by using techniques based on curvature feature extraction or ridge direction change detection. The aim is to enhance feature diversity by utilizing the stability of local fingerprint details, especially to meet the challenge of information sparseness in small-area images.

[0076] The scheme in this application obtains the bifurcation points and endpoints of the fingerprint ridges as first feature points through an initial search, and then triggers adaptive branching logic based on a feature point quantity threshold. When the number of first feature points is sufficient, quality screening is performed first to ensure feature point accuracy and prevent inferior points from interfering with subsequent matching. When the number of first feature points is insufficient, the point with the maximum curvature of the ridge is actively introduced as a second feature point to supplement it, using the sensitivity of curvature points to morphological changes to compensate for the lack of traditional features. This threshold-driven dynamic construction mechanism enables the feature point set to be intelligently adjusted according to the actual image conditions, ensuring that sufficient information and robustness are always maintained in small-area fingerprint scenarios, thereby effectively supporting the accurate extraction and matching of local fingerprint image features.

[0077] As a preferred embodiment, the solution of this application is implemented as follows: When extracting ridge feature points from the fingerprint image to be identified, the system first performs a first feature point search. If the number of first feature points is detected to be lower than a preset threshold, a second feature point search process is initiated. By analyzing the curvature change characteristics of the fingerprint ridges, the maximum curvature point is located, and the two types of feature points are fused to form a feature point set. If the number of first feature points is sufficient, each point is evaluated for quality, and only feature points with high contrast and clear structure in the neighboring pixels are retained. When this embodiment is run on a small-area fingerprint sensor of a mobile device, it can dynamically adjust the source of feature points according to the real-time image quality, avoiding feature loss problems caused by seasonal changes or skin condition.

[0078] Through the above solution, the embodiments of this application effectively solve the problem of reduced feature set information caused by insufficient number of feature points in small-area fingerprint images, ensuring the reliability and diversity of feature point set in sparse scenarios, thereby significantly improving the stability and accuracy of fingerprint recognition process.

[0079] In some embodiments, searching for a first feature point in the fingerprint image to be identified includes:

[0080] Fingerprint ridge features are extracted from the fingerprint image to be identified to obtain a ridge feature map;

[0081] Calculate the total color difference of the neighboring pixels of the target pixel; the target pixel is the pixel that constitutes the fingerprint ridge in the ridge feature map, and the total color difference of the neighboring pixels is the sum of the pixel color differences between each pair of adjacent pixels of the target pixel;

[0082] Based on the total color difference of neighboring pixels, the bifurcation points and endpoints of the fingerprint ridges in the ridge feature map are determined to obtain the first feature point.

[0083] Fingerprint ridge feature extraction refers to the process of extracting a clear and coherent ridge structure representation from the original fingerprint image. This can be achieved using image enhancement techniques based on orientation field estimation or adaptive filtering methods, with the aim of transforming blurred ridges into structured features to suppress background noise. A ridge feature map is a binary image representing the distribution of fingerprint ridges. It can be obtained through image thinning and connected component processing, with the aim of providing a stable structural basis for feature point localization.

[0084] The target pixel refers to the pixel in the ridge feature map that actually constitutes the fingerprint ridge. It can be determined by pixel value thresholding or morphological operations to limit the calculation range and improve processing efficiency. The total color difference of neighboring pixels refers to the cumulative quantized value of the color difference between adjacent pixel pairs in the neighborhood of the target pixel. It can be achieved by calculating the sum of the absolute values ​​of the grayscale differences of all adjacent pixel pairs in the neighborhood of the target pixel, with the aim of capturing the global characteristics of abrupt changes in ridge direction.

[0085] The solution in this application extracts fingerprint ridge features from the fingerprint image to be identified, converting it into a clear ridge feature map. This effectively suppresses noise interference in small-area images while preserving ridge structure details. Subsequently, the total color difference of the target pixel's neighboring pixels is calculated. This total color difference is significantly higher in regions where the ridge direction changes drastically (such as bifurcation points or endpoints) than in smooth regions, thus avoiding the sensitivity to noise inherent in traditional local gradient detection. Finally, the feature point location is determined based on the total color difference. Utilizing the high response of color difference accumulation to ridge discontinuities, key feature points are stably extracted from low-quality images. This synergistic mechanism of structured feature extraction and global color difference quantization ensures that feature point detection is unaffected by seasonal peeling or changes in moisture levels, providing reliable input for subsequent matching.

[0086] As a preferred embodiment, the solution of this application is specifically implemented as follows: fingerprint ridge feature extraction adopts orientation field estimation combined with image enhancement technology, and the original image is filtered and binarized to obtain ridge feature map; the target pixel is selected as the pixel with a value of 1 in the ridge feature map; when calculating the total color difference of neighboring pixels, the sum of the absolute values ​​of the gray level differences of all adjacent pixel pairs in the neighborhood centered on the target pixel is calculated; when the sum exceeds a preset threshold, it is determined as a bifurcation point or endpoint.

[0087] Through the above scheme, this application can reliably extract the bifurcation points and endpoints of fingerprint ridges in small-area fingerprint images, reducing false detections and missed detections caused by noise and loss of details, thereby improving the accuracy and reliability of subsequent feature matching.

[0088] In a specific embodiment, fingerprint ridge feature extraction of the fingerprint image to be identified specifically includes performing image enhancement processing, Gabor filtering enhancement processing, binarization processing, and feature thinning processing on the fingerprint image to be identified in sequence, and finally obtaining the ridge feature map.

[0089] In a specific embodiment, calculating the total color difference of the target pixel's neighboring pixels specifically includes: determining whether the pixel is a target pixel based on the pixel color of the pixel in the ridge feature map and the pixel colors of its eight neighboring pixels; and calculating the total color difference of the target pixel's neighboring pixels after determining it to be a target pixel. The conditions for determining that the pixel is a target pixel are: the pixel's binarized pixel value is 1; not all of the pixel's off-diagonal neighboring pixels have binarized pixel values ​​of 1; at least two of the pixel's neighboring pixels have binarized pixel values ​​of 1; the pixel's octet connectivity is 1; and the pixel's octet connectivity is 1 when the binarized pixel value of any off-diagonal neighboring pixel is 0. The formula for calculating the pixel's octet connectivity is:

[0090] ,

[0091] ,

[0092] in, Let P be the number of octet connections. , and All are neighboring pixels of pixel P, where k is a natural number, k∈[0,3];

[0093] The formula for calculating the total color difference of the target pixel's neighboring pixels is:

[0094] ,

[0095] ,

[0096] in, For target pixel The total color difference of neighboring pixels, and All are target pixels The neighboring pixels, j is a natural number, j∈[0,7], when At time 2, the target pixel The endpoints of the fingerprint ridges, At 6 o'clock, the target pixel This is the bifurcation point of the fingerprint ridge.

[0097] In some embodiments, searching for a second feature point in the fingerprint image to be identified includes:

[0098] The orientation feature map is obtained by extracting orientation features from the fingerprint image to be identified.

[0099] Curvature feature extraction is performed on the orientation feature map to obtain the curvature feature map;

[0100] Gaussian smoothing is applied to the curvature feature map, and the pixel with the maximum pixel value after Gaussian smoothing is searched to obtain the second feature point.

[0101] Orientation feature extraction refers to determining the local orientation information of fingerprint ridges. This can be achieved using Gabor filters or histograms of oriented gradients, with the aim of providing stable ridge orientation information and reducing noise interference in the original image. An orientation feature map can be understood as a two-dimensional image representing the orientation of fingerprint ridges, specifically storing the orientation angle value of each pixel. Its purpose is to provide reliable basic data for subsequent curvature calculations.

[0102] Curvature feature extraction refers to quantifying the curvature of fingerprint ridges. This can be achieved by calculating the rate of change of direction or the magnitude of the directional gradient based on the directional feature map, with the aim of highlighting the areas where the ridge morphology changes most significantly. The curvature feature map is an image that stores curvature values, specifically the distribution of curvature magnitude for each pixel, with the purpose of identifying the region of maximum curvature.

[0103] Gaussian smoothing refers to applying a Gaussian filter to smooth an image. This can be achieved using Gaussian kernels with different standard deviations or bilateral filters, with the aim of filtering out spurious peaks caused by noise. Searching for the maximum pixel value involves locating the point with the highest pixel value in the smoothed curvature feature map. This can be achieved using local maximum search algorithms or non-maximum suppression techniques, with the aim of accurately obtaining the second feature point.

[0104] The solution in this application first extracts directional features from the fingerprint image to be identified, using ridge direction information as a basis. This avoids interference from local noise when operating directly on the original image, ensuring accurate capture of the overall direction of the fingerprint ridges even with poor image quality. Curvature features are then extracted based on the directional feature map, precisely quantifying the degree of ridge curvature through directional information, highlighting the region of maximum curvature, and focusing feature point localization on the location of the most significant ridge morphological change. After Gaussian smoothing the curvature feature map, the pixel with the maximum pixel value is searched. Gaussian smoothing effectively filters out false peaks caused by noise, making the curvature feature distribution smoother and ensuring that the searched points truly reflect the location of the maximum ridge curvature, avoiding interference from environmental factors such as changes in finger dryness or wetness on feature point extraction.

[0105] As a preferred embodiment, the solution of this application is implemented as follows: Orientation feature extraction uses a Gabor filter bank to convolve the fingerprint image to obtain an orientation feature map. Curvature feature extraction calculates the directional gradient magnitude of each pixel based on the orientation feature map to obtain a curvature feature map. Gaussian smoothing applies a standard Gaussian kernel to filter the curvature feature map; finally, the global maximum point is searched in the smoothed curvature feature map as the second feature point.

[0106] Through the above solution, the embodiments of this application effectively solve the problem of unstable extraction of the second feature point in small-area fingerprint images, improve the stability and repeatability of feature points under noise and environmental changes, thereby improving the accuracy of fingerprint recognition.

[0107] In a specific embodiment, extracting directional features from the fingerprint image to be identified specifically includes calculating the Sobel gradient field of the fingerprint image to be identified:

[0108] ,

[0109] ,

[0110] in, and All are Sobel gradient fields. This is the enhanced feature map obtained by performing image enhancement processing on the fingerprint image to be identified.

[0111] Then, the polarity is eliminated using the double-angle formula, and a region (in this paper's method, the region has a radius of...) is used. a square, The value is 4) The averaging method is used to calculate the direction of each pixel. :

[0112] ,

[0113] ,

[0114] ,

[0115] in, The gradient in the y-direction is a double angle. The gradient in the x-direction is a double angle. This is a directional feature map.

[0116] In a specific embodiment, the expression for curvature feature extraction from the orientation feature map is:

[0117] ,

[0118] in, This is a curvature feature map. To evaluate the radius.

[0119] In one specific embodiment, the expression for searching the pixel with the maximum pixel value is:

[0120]

[0121] in, The direction of the point of maximum curvature The x-coordinate of the point of maximum curvature The ordinate of the point of maximum curvature is... For the smooth radius, Represents the imaginary unit. This indicates taking the principal argument of a complex number.

[0122] In some embodiments of this application, image feature extraction based on feature points in a feature point set is proposed to obtain local fingerprint image features. However, in its implementation, the screenshot operation does not fully incorporate the directional information of the feature points for dynamic alignment, resulting in the screenshot area possibly containing invalid regions (such as background noise or non-fingerprint areas) or directional deviations. This leads to inaccurate local feature extraction and susceptibility to the sparse distribution of feature points in small fingerprint areas, thereby reducing the reliability of feature matching and overall recognition performance. Therefore, this application further proposes a method for image feature extraction from the fingerprint image to be identified.

[0123] See Figure 4 In one embodiment, the method for extracting image features from the fingerprint image to be identified includes, but is not limited to, steps S401 to S402.

[0124] Step S401: Using the feature points in the feature point set as the center, take a screenshot of the fingerprint image to be identified to obtain multiple original feature point images.

[0125] Step S402: Extract image features from the original feature point map to obtain local fingerprint image features.

[0126] Centering the screenshot on a feature point within a feature point set refers to dynamically adjusting the screenshot area to strictly focus on the local context structure of the feature point. This can be achieved by determining a fixed-size window based on the feature point coordinates or by adaptively adjusting the window shape according to the ridge direction. The goal is to ensure that the screenshot area accurately covers the key fingerprint ridges around the feature point, avoiding background noise or non-fingerprint areas that may be introduced by traditional fixed-area screenshots. Image feature extraction from the original feature point map refers to extracting discriminative feature representations from the aligned local image. This can be achieved using feature description algorithms such as SIFT, SURF, or Akaze. The aim is to isolate global image noise interference, focusing the feature extraction process on effective ridge structures, thereby improving the stability and repeatability of the features.

[0127] The solution in this application performs screenshot operations centered on feature points within a feature point set, ensuring that each screenshot area strictly corresponds to the spatial location of the feature points, effectively isolating global noise interference. Simultaneously, multiple original feature point maps are generated to construct a multi-view local feature set, which not only enhances information redundancy to compensate for the insufficient overall information content of small-area fingerprints but also disperses the risk of single-point errors. Finally, image feature extraction is performed on the original feature point maps to ensure that each local feature originates from a precisely aligned region, maintaining high consistency in feature position and orientation, thus forming a complete technical chain and significantly improving the robustness of small-area fingerprint recognition in complex environments.

[0128] As a specific implementation method, the solution of this application embodiment is implemented as follows: For each feature point in the fingerprint image to be identified, the execution subject determines a square screenshot area centered on the coordinates of the feature point. The size of this area is sufficient to cover the key ridge structure around the feature point. Subsequently, the SIFT algorithm is used to perform feature description on each screenshot area, generating a 128-dimensional feature vector as a local fingerprint image feature. This feature vector can effectively characterize the texture and direction information of the local ridges.

[0129] In a specific embodiment, image feature extraction from the original feature point map includes inputting the original feature point map into a preset image feature extraction network to obtain corresponding local fingerprint image features. This image feature extraction network may include convolutional layers, depthwise separable convolutional layers, normalization layers, nonlinear activation function layers, max pooling layers, and global pooling layers, with connections between layers including feedforward connections and residual connections.

[0130] Through the above solution, the embodiments of this application effectively avoid the problem of the screenshot area containing invalid areas or directional deviation, improve the accuracy of local feature extraction, enhance the reliability of feature matching, and thus improve the overall performance of small-area fingerprint recognition in complex scenarios such as changes in dryness and wetness or slight peeling.

[0131] In some embodiments, a screenshot of the fingerprint image to be identified is taken, centered on a feature point in the feature point set, including:

[0132] Based on the position and orientation information of the feature points in the feature point set, a rigid transformation is performed on the fingerprint image to be identified to obtain a transformed feature map centered on the feature points in the feature point set.

[0133] The transformed feature map is truncated and invalid regions are labeled to obtain the original feature point map.

[0134] Rigid transformation refers to a geometric transformation operation that keeps the distance between points in an image constant. It can be achieved by a combination of rotation and translation, and can be calculated using an affine transformation matrix or an Euclidean transformation. Its purpose is to correct the directional deviation of fingerprint ridges and achieve geometric center alignment of feature points.

[0135] Invalid region labeling refers to the process of identifying and marking unreliable or noisy regions in an image. It can be achieved using methods based on local pixel gradient analysis or edge continuity detection. Its purpose is to actively shield invalid pixels caused by image edge distortion or changes in skin condition, thereby improving the purity of local features.

[0136] The scheme of this application first performs a rigid transformation on the fingerprint image to be identified based on the position and orientation information of the feature points in the feature point set. The position information aligns the geometric center of the feature points to standard coordinates, while the orientation information corrects the orientation deviation of the fingerprint ridges, generating a standardized transformed feature map. Subsequently, the transformed feature map is truncated and invalid regions are labeled. The truncating operation relies on the corrected image to precisely define the local area range, while the invalid region labeling actively identifies and masks interfering parts by analyzing pixel reliability, thereby eliminating invalid pixels introduced by finger tilt or rotation, ensuring that the original feature point map focuses on the valid ridge structure. This process, driven by both position and orientation information, avoids the feature point offset problem in traditional truncating methods and provides standardized input for subsequent feature extraction.

[0137] As a specific embodiment, the solution of this application is implemented as follows: When performing a rigid transformation on the fingerprint image to be identified, the coordinate offset and orientation angle difference of the feature points are calculated, and the corresponding rotation and translation transformations are applied to center the feature points and standardize their orientation. Subsequently, a region of fixed shape is cropped with the feature points as the center, and edge distortion regions are identified by analyzing local pixel gradient changes and binarized to obtain the original feature point map of the focused effective ridge.

[0138] Through the above-described solution, the embodiments of this application effectively avoid the problem of feature points deviating from the center position or having inconsistent directions, reduce the impact of invalid areas and noise interference on local image feature extraction, improve the accuracy and robustness of feature matching, and thus significantly improve the stability of recognition performance in small-area fingerprint recognition scenarios.

[0139] In one specific embodiment, after the fingerprint image to be identified is rigidly transformed to a reference frame, the feature points in the feature point set are transformed to the center, and the orientation of the feature points is horizontal to the right in the reference frame. The expression for the rigid transformation of the fingerprint image to be identified is:

[0140] ,

[0141] in, To transform the feature map, v represents the location information of the feature points. For the orientation information of feature points, This is to transform the side length of the feature map.

[0142] In a specific embodiment, the process of truncating the transformed feature map and marking invalid regions includes, after obtaining the transformed feature map, truncating the transformed feature map, and using bilinear interpolation, interpolating regions outside the truncated transformed feature map with a value of 255 (indicating invalid regions) to obtain the original feature point map.

[0143] In some embodiments described above, a method for matching local fingerprint image features with preset target fingerprint image features to generate fingerprint recognition results is proposed. However, in practice, due to the limited number of feature points in a small fingerprint area and its susceptibility to environmental factors such as season, peeling, and dryness / wetness, simple feature matching cannot effectively distinguish matching quality, leading to a decrease in recognition accuracy. Therefore, embodiments of this application further propose a method for matching local fingerprint image features with preset target fingerprint image features.

[0144] See Figure 5 In one embodiment, the method for matching local fingerprint image features with preset target fingerprint image features includes, but is not limited to, steps S501 to S502.

[0145] Step S501: Calculate the feature matching degree between the local fingerprint image features and the target fingerprint image features.

[0146] Step S502: Based on the feature matching degree, the local fingerprint image features generated based on the first feature point are matched with the target fingerprint image features generated based on the first feature point, and / or the local fingerprint image features generated based on the second feature point are matched with the target fingerprint image features generated based on the second feature point, and a fingerprint recognition result of the fingerprint image to be recognized is generated based on the matching result.

[0147] Feature matching degree refers to an indicator used to quantify the similarity between local fingerprint image features and target fingerprint image features. It can be implemented using similarity calculations based on Euclidean distance, evaluation methods based on cosine similarity, or matching scores based on deep learning models. Its purpose is to provide a quantifiable standard to distinguish between different levels of matching quality. A feature matching degree-based matching mechanism can be understood as dynamically deciding whether to include matching results of specific types of feature points based on the numerical range or threshold setting of the feature matching degree. This can be achieved using hierarchical threshold determination strategies, adaptive weight allocation mechanisms, or priority matching rules. Its aim is to ensure the matching priority of key feature points while also considering the contribution of information-rich feature points, thereby optimizing the overall matching reliability under the condition of a limited number of feature points.

[0148] In some embodiments, calculating the feature matching degree between local fingerprint image features and target fingerprint image features includes:

[0149] Based on the rigid transformation mapping relationship between local fingerprint image features and target fingerprint image features, the consistency of feature point position dimension and feature point orientation dimension between local fingerprint image features and target fingerprint image features is verified, and the consistency verification result is obtained.

[0150] When the consistency verification result meets the preset consistency conditions, the feature matching degree is calculated based on the average similarity, highest similarity, and lowest similarity between the local fingerprint image features and the target fingerprint image features.

[0151] Rigid transformation mapping refers to a transformation model used to correct geometric differences between images. It can be implemented using affine transformation or Euclidean transformation, with the aim of compensating for slight displacements during fingerprint acquisition. Consistency verification of feature point position dimension and feature point orientation dimension refers to verifying the correspondence between feature points in spatial coordinates and orientation angles. It can be implemented using distance threshold comparison or angle difference calculation, with the aim of ensuring the reliability of feature point geometric relationships.

[0152] The consistency verification result refers to the judgment value output by the consistency verification mechanism. It can be represented by a Boolean value or a confidence score, and its purpose is to provide a reliable basis for the matching process. The preset consistency condition refers to the threshold standard used to judge whether the consistency verification result is valid. It can be implemented by using a fixed threshold or dynamically adjusting the threshold, in order to adapt to the matching needs in different environments.

[0153] The solution in this application first performs geometric correction based on the rigid transformation mapping relationship between local fingerprint image features and target fingerprint image features. This effectively compensates for the slight rotation or translation displacement of the finger commonly encountered during fingerprint acquisition, ensuring that feature points maintain their geometric correspondence after spatial transformation. Subsequently, the consistency of the feature point position and orientation dimensions is verified. This dual-dimensional constraint strengthens the reliability of the spatial relationship of feature points, significantly reducing the interference of noise points or outliers on the matching process. Feature matching degree is calculated based on average similarity, highest similarity, and lowest similarity only when the consistency verification result meets preset conditions. This multi-index fusion strategy evaluates the overall matching quality through average similarity, highlights the contribution of the optimal matching point through highest similarity, and suppresses the influence of outliers through lowest similarity, forming a comprehensive coverage of similarity evaluation. This adapts to scenarios with a limited number of fingerprint feature points in a small area, effectively alleviating the feature stability problem caused by seasonal changes or fluctuations in skin condition.

[0154] As a preferred embodiment, the specific implementation of the scheme in this application embodiment is as follows: In the fingerprint recognition module of the mobile terminal, the calculation of feature matching degree is performed by the application processor. The processor adopts the central processing unit of ARM architecture, and realizes the calculation of rigid transformation mapping relationship and consistency verification by calling the OpenCV image processing library, and calculates the similarity index using SIFT feature descriptor. The rigid transformation mapping relationship is determined by the homography matrix generated by feature point matching, and the consistency verification is judged by the position deviation threshold and the direction deviation threshold.

[0155] Through the above-mentioned solutions, the embodiments of this application effectively solve the problem of decreased matching accuracy caused by limited feature information in small-area fingerprint recognition. The dual verification mechanism of feature point position and orientation dimension avoids false matching. The multi-similarity index fusion strategy significantly reduces the impact of image noise and skin condition changes on the matching results, thereby improving the reliability and accuracy of fingerprint recognition in complex environments.

[0156] In one specific embodiment, local fingerprint image features and target fingerprint image features The expression for the rigid transformation mapping relationship between them is:

[0157] ,

[0158] in, This represents the rigid transformation mapping relationship between the a-th local fingerprint image feature and the b-th target fingerprint image feature. , , and These represent the horizontal pixel coordinates, vertical pixel coordinates, orientation, and feature vector of the feature points corresponding to the local fingerprint image features in the fingerprint image to be identified. , , and These represent the horizontal pixel coordinates, vertical pixel coordinates, direction, and feature vector of the feature points corresponding to the features in the target fingerprint image.

[0159] use Functions to remove local fingerprint image features It checks all other feature points besides the corresponding feature point, and then verifies whether the transformed feature points are consistent with their optimal pairing point in terms of coordinate position and angle. For example, local fingerprint image features. The optimal pairing of target fingerprint image features is , go through Transformed coordinates and angle for:

[0160] ,

[0161] ,

[0162] If both of the following conditions are met:

[0163] ,

[0164] ,

[0165] This indicates local fingerprint image features. Optimal pairing and It is about consistency. In this embodiment, the threshold... , .

[0166] In There is at least one consistent optimal pairing, that is... and Assumptions and Consistent optimal pairings include ( ) , calculate all those with The average, highest, and lowest similarities between local fingerprint image features and target fingerprint image features that meet the consistency condition are calculated. These average, highest, and lowest similarities are then weighted and summed to obtain the corresponding feature matching degree. After obtaining the feature matching degrees between each local fingerprint image feature and the target fingerprint image feature, the feature matching degree with the highest matching degree is taken as the image matching degree between the fingerprint image to be identified and the target fingerprint image, thus generating the fingerprint recognition result of the fingerprint image to be identified.

[0167] Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application.

[0168] The following reference Figure 6 To describe an electronic device 600 according to such an embodiment of the present disclosure. Figure 6 The electronic device 600 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0169] like Figure 6 As shown, the electronic device 600 is presented in the form of a general-purpose computing device. The components of the electronic device 600 may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different system components (including storage unit 620 and processing unit 610), a display unit 640, etc.

[0170] The storage unit stores program code, which can be executed by the processing unit 610, causing the processing unit 610 to perform the steps described in the above-described small-area fingerprint recognition method section of this specification according to various exemplary embodiments of this disclosure.

[0171] Storage unit 620 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 6201 and / or cache memory 6202, and may further include a read-only memory (ROM) 6203.

[0172] Storage unit 620 may also include a program / utility 6204 having a set (at least one) program module 6205, such program module 6205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0173] Bus 630 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0174] Electronic device 600 can also communicate with one or more external devices 600' (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 600, and / or with any device that enables electronic device 600 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 650. Furthermore, electronic device 600 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 660. Network adapter 660 can communicate with other modules of electronic device 600 via bus 630. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0175] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0176] The small-area fingerprint recognition method, device, and medium provided in this application construct a feature point set by fusing stable ridge topological features with highly discriminative curvature features, and implement localized feature processing and matching strategies. This effectively addresses the accuracy decline problem caused by insufficient traditional feature information in small-area fingerprint recognition. The diversified design of the feature point set enhances information representation within a limited area, the local feature extraction strategy reduces the negative impact of environmental factors on feature stability, and the feature matching process strengthens tolerance to deformation and noise through targeted comparison of local features. Thus, it maintains high-reliability fingerprint recognition performance under limited sensor area conditions, avoids the performance fluctuation problem caused by existing machine vision methods relying on pixel-level changes, and improves the recognition accuracy of small-area fingerprint recognition.

[0177] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, or network device, etc.) to execute the methods described above according to the embodiments of this disclosure.

[0178] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0179] Computer-readable storage media may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0180] Those skilled in the art will understand that the above modules can be distributed in the device as described in the embodiments, or they can be modified to be uniquely different from one or more devices in this embodiment. The modules in the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0181] Exemplary embodiments of this disclosure have been specifically shown and described above. It should be understood that this disclosure is not limited to the detailed structures, arrangements, or implementations described herein; rather, this disclosure is intended to cover various modifications and equivalent arrangements contained within the spirit and scope of the appended claims.

Claims

1. A method for small-area fingerprint recognition, characterized in that, include: Acquire the fingerprint image to be identified; The fingerprint image to be identified is subjected to ridge feature point extraction to obtain a feature point set; The feature point set includes a first feature point or includes the first feature point and a second feature point. The first feature point is the bifurcation point and / or endpoint of the fingerprint ridge, and the second feature point is the point of maximum curvature of the fingerprint ridge. Based on the feature points in the feature point set, image feature extraction is performed on the fingerprint image to be identified to obtain several corresponding local fingerprint image features; The local fingerprint image features are matched with the preset target fingerprint image features, and the fingerprint recognition result of the fingerprint image to be recognized is generated based on the feature matching result. The step of extracting ridge feature points from the fingerprint image to be identified includes: Search for the first feature point in the fingerprint image to be identified; Determine whether the number of the first feature points exceeds the feature point number threshold; If the number of features exceeds the limit, a feature point quality evaluation is performed on the first feature point, and the first feature points that meet the feature point quality evaluation conditions are selected to form the feature point set. If the limit is not exceeded, search for the second feature point in the fingerprint image to be identified, and form the feature point set with the first feature point and the second feature point; The step of extracting image features from the fingerprint image to be identified based on the feature points in the feature point set includes: Using the feature points in the feature point set as the center, the fingerprint image to be identified is captured to obtain multiple original feature point images; Image features are extracted from the original feature point map to obtain the local fingerprint image features.

2. The small-area fingerprint recognition method according to claim 1, characterized in that, The search for the first feature point in the fingerprint image to be identified includes: Fingerprint ridge features are extracted from the fingerprint image to be identified to obtain a ridge feature map; Calculate the total color difference of the neighboring pixels of the target pixel; the target pixel is the pixel that constitutes the fingerprint ridge in the ridge feature map, and the total color difference of the neighboring pixels is the sum of the pixel color differences between each pair of adjacent neighboring pixels of the target pixel; Based on the total color difference of the neighboring pixels, the bifurcation point and endpoint of the fingerprint ridge in the ridge feature map are determined to obtain the first feature point.

3. The small-area fingerprint recognition method according to claim 1, characterized in that, The search for the second feature point in the fingerprint image to be identified includes: The fingerprint image to be identified is subjected to directional feature extraction to obtain a directional feature map; Curvature feature extraction is performed on the directional feature map to obtain a curvature feature map; The curvature feature map is Gaussian smoothed, and the pixel with the maximum pixel value is searched after Gaussian smoothing to obtain the second feature point.

4. The small-area fingerprint recognition method according to claim 1, characterized in that, The step of taking a screenshot of the fingerprint image to be identified, centered on the feature points in the feature point set, includes: Based on the position and orientation information of the feature points in the feature point set, a rigid transformation is performed on the fingerprint image to be identified to obtain a transformed feature map centered on the feature points in the feature point set. The transformed feature map is truncated and invalid regions are labeled to obtain the original feature point map.

5. The small-area fingerprint recognition method according to claim 1, characterized in that, The step of matching the local fingerprint image features with the preset target fingerprint image features includes: Calculate the feature matching degree between the local fingerprint image features and the target fingerprint image features; Based on the feature matching degree, the local fingerprint image features generated based on the first feature point are matched with the target fingerprint image features generated based on the first feature point, or the local fingerprint image features generated based on the first feature point are matched with the target fingerprint image features generated based on the first feature point, and the local fingerprint image features generated based on the second feature point are matched with the target fingerprint image features generated based on the second feature point. The fingerprint recognition result of the fingerprint image to be identified is generated based on the matching result.

6. The small-area fingerprint recognition method according to claim 5, characterized in that, The calculation of the feature matching degree between the local fingerprint image features and the target fingerprint image features includes: Based on the rigid transformation mapping relationship between the local fingerprint image features and the target fingerprint image features, the consistency of the feature point position dimension and feature point direction dimension between the local fingerprint image features and the target fingerprint image features is verified to obtain the consistency verification result. When the consistency verification result meets the preset consistency conditions, the feature matching degree is calculated based on the average similarity, highest similarity, and lowest similarity between the local fingerprint image features and the target fingerprint image features.

7. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the small-area fingerprint recognition method according to any one of claims 1 to 6.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the small-area fingerprint recognition method according to any one of claims 1 to 6.

Images