Small area fingerprint image matching method, device, equipment and storage medium

By extracting and aligning the ridge point and line features of small-area fingerprint images, and combining the overall position deviation and alignment overlap, multiple matching is performed, the problem of low matching accuracy of small-area fingerprint images is solved and the accuracy of fingerprint recognition is improved.

CN118522041BActive Publication Date: 2024-11-12SHENZHEN MAGIC INFORMATION TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410973607.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-12
Estimated Expiration
2044-07-19

AI Technical Summary

Technical Problem

Due to the small-area fingerprint image, the small-area fingerprint image has insufficient fingerprint details, which affects the accuracy of fingerprint feature comparison and thus affects the accuracy of fingerprint recognition.

Method used

By obtaining the fingerprint image to be matched, preprocessing is performed to extract the ridgeline map, then extracting the ridge point features and line body features, performing ridge line alignment and point matching, combining the overall position deviation and alignment overlap of the ridge line points in the alignment area. , perform multiple matching and comparison.

Benefits of technology

The matching accuracy of small-area fingerprint images is improved, and the defects of single fingerprint image matching mechanism and single feature types are overcome, achieving more accurate fingerprint recognition.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118522041B_ABST
    Figure CN118522041B_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of image processing, and discloses a small-area fingerprint image matching method, device, equipment and storage medium. The method comprises: obtaining at least two fingerprint images to be matched; preprocessing the fingerprint images to be matched to obtain a ridge map containing ridges; extracting ridge point features of the ridges to obtain ridge point features; extracting body features of the ridges to obtain ridge body features; performing ridge alignment operations on the two ridge maps based on the ridge point features and ridge body features to obtain an alignment area; performing ridge point matching operations on the two ridge maps based on the overall position deviation or local position deviation of the ridge points in the alignment area to obtain a ridge point matching result; obtaining a matching result of the two fingerprint images to be matched based on the ridge point matching result. The embodiments of the present application can improve the matching accuracy of small-area fingerprint images.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a small-area fingerprint image matching method, device, equipment and storage medium. Background Art

[0002] Biometric technology is an important means to achieve identity recognition by comparing people's unique biological characteristics. Among them, fingerprint recognition technology is the mainstream recognition technology in the market.

[0003] In the related technologies, fingerprint recognition technologies include optical method, capacitance method, ultrasonic method, etc. The optical method is widely used in imaging, sensing, storage and other fields due to its fast response speed. The main steps of optical fingerprint recognition technology include fingerprint collection, preprocessing of the collected fingerprint pattern (such as grayscale processing, filtering enhancement, etc.), feature point extraction of the preprocessed fingerprint, and comparison of the extracted feature points to achieve fingerprint recognition.

[0004] However, small-area fingerprint images have a small image area, resulting in insufficient fingerprint detail information, which affects the accuracy of fingerprint feature comparison and thus affects the accuracy of fingerprint recognition. Summary of the invention

[0005] The purpose of this application is to provide a small-area fingerprint image matching method, device, equipment and storage medium, aiming to improve the matching accuracy of small-area fingerprint images.

[0006] The present application embodiment provides a small-area fingerprint image matching method, including:

[0007] Obtain at least two fingerprint images to be matched;

[0008] Preprocessing the fingerprint image to be matched to obtain a ridge line map containing ridge lines;

[0009] Extracting ridge point features from the ridge to obtain ridge point features;

[0010] Extracting line features of the ridge line to obtain ridge line features;

[0011] According to the ridge point features and the ridge body features, a ridge alignment operation is performed on the two ridge graphs to obtain an alignment area; the alignment area is an area where the two ridge graphs partially overlap;

[0012] According to the overall position deviation of the ridge points in the alignment area, or the local position deviation of the local position of the ridge points in the alignment area with the highest degree of alignment overlap when the overall position deviation of the ridge points in the alignment area exceeds a preset deviation range, a ridge point matching operation is performed on the two ridge line graphs to obtain a ridge point matching result;

[0013] According to the ridge point matching result, the matching result of the two fingerprint images to be matched is obtained.

[0014] In some embodiments, the ridge point features include ridge point category features and ridge point association features, and the ridge point feature extraction for the ridge line to obtain the ridge point features includes:

[0015] Extracting ridge points in the ridge map;

[0016] Classifying the ridge points according to the neighborhood colors of the ridge points and the positions in the ridge map to obtain the ridge point category features;

[0017] According to the ridge point category feature, the ridge line is tracked to obtain a tracking result, and according to the tracking result, the ridge line and the ridge point are associated to obtain the ridge point association feature.

[0018] In some embodiments, the ridge line features include line length features, curvature features, line width features, and line type features, and the ridge line features are extracted to obtain the ridge line features, including:

[0019] Performing Freeman chain code encoding on the ridge to obtain a ridge chain code representation;

[0020] Performing a mapping accumulation operation on the ridge chain code representation to obtain the line length feature;

[0021] Calculating the curvature of the ridge line according to the relative positions of the ridge line points on the ridge line to obtain the curvature feature;

[0022] Taking the sampling points on the ridge line as a reference, performing line width detection on the ridge line in a direction perpendicular to the ridge line to obtain the line width feature;

[0023] According to the target ridge point feature, the ridge line is classified to obtain the line type feature; the target ridge point feature is the ridge point feature of the ridge point located at the beginning and end positions of the ridge line;

[0024] When the line feature characterizes that the ridge line is a target ridge line category, the ridge line is sampled at intervals, and a corresponding ridge line descriptor is generated according to the interval sampling result.

[0025] In some embodiments, the ridge line feature includes a line length feature, a curvature feature, a line width feature, and a line type feature, and the ridge line alignment operation is performed on the two ridge line graphs according to the ridge line point feature and the ridge line feature to obtain an alignment area, including:

[0026] When the ridge line meets the first alignment condition, similarity matching and rigid transformation matching are performed on the ridge line according to the line length feature, and several pairs of the ridge lines with the highest matching degree are determined to obtain a first matching result; the first alignment condition is that the starting ridge line point and / or the ending ridge line point of the ridge line is formed by a non-linear interruption or a linear bifurcation;

[0027] When the ridge line meets the second alignment condition, the ridge line is positioned at the inflection point according to the curvature feature, and the ridge line is matched with the line coincidence and rigid transformation by using the inflection point obtained by positioning, and several pairs of ridge lines with the highest matching degree are determined to obtain a second matching result; the second alignment condition is that the maximum curvature of the ridge line is greater than a preset maximum curvature threshold;

[0028] When the ridge lines do not meet the first alignment condition and the second alignment condition, rigid transformation matching and transformation angle estimation are performed on the ridge lines according to the curvature feature and the line width feature, and line coincidence matching is performed after the ridge lines are rotated to the estimated transformation angle, and several pairs of ridge lines with the highest matching degree are determined to obtain a third matching result;

[0029] According to the first matching result, the second matching result or the third matching result, the two ridge line graphs are aligned to form the alignment area.

[0030] In some embodiments, performing similarity matching and rigid transformation matching on the ridge lines according to the line length features, determining several pairs of ridge lines with the highest matching degree, and obtaining a first matching result includes:

[0031] Calculating the line length deviation between the ridge lines according to the line length feature;

[0032] When the line length deviation is less than a preset line length deviation threshold, similarity matching is performed on the ridge descriptors of the ridge lines to obtain descriptor similarity;

[0033] When the similarity of the descriptors is greater than a preset similarity threshold, rigid transformation matching is performed on the ridge line, and the rotation matrix and translation parameters of the ridge line are solved by the least square method;

[0034] When the rotation matrix is ​​smaller than a preset rotation threshold and the translation parameter is smaller than a preset translation threshold, selecting a plurality of pairs of ridge lines whose rotation matrix is ​​closest to the preset rotation threshold and whose translation parameter is closest to the preset translation threshold as the first matching results;

[0035] And / or locating the inflection point of the ridge line according to the curvature feature, performing line coincidence matching and rigid transformation matching on the ridge line using the inflection point obtained by locating, determining several pairs of ridge lines with the highest matching degree, and obtaining a second matching result, including:

[0036] Determining the maximum curvature of the ridge line according to the curvature feature;

[0037] When the maximum curvature is greater than a preset maximum curvature threshold, calculating the maximum curvature deviation between the ridge lines;

[0038] When the maximum curvature deviation is less than a preset curvature deviation threshold, a chain code derivative integration method is used to locate the inflection point of the ridge line, and the inflection point is located;

[0039] Calculating the line coincidence matching degree of the ridge line according to the number and relative angle of the inflection points;

[0040] When the line coincidence matching degree is less than a preset line coincidence threshold, rigid transformation matching is performed on the ridge lines to determine a number of pairs of ridge lines with the highest matching degree, and obtain a second matching result;

[0041] And / or the step of performing rigid transformation matching and transformation angle estimation on the ridge line according to the curvature feature and the line width feature, rotating the ridge line to the estimated transformation angle and then performing line coincidence matching, determining several pairs of the ridge lines with the highest matching degree, and obtaining a third matching result, including:

[0042] Calculating the average curvature and average line width of the ridge line according to the curvature feature and the line width feature;

[0043] When the average curvature is greater than a preset average curvature threshold and the average line width is greater than a preset average line width threshold, rigid transformation matching is performed on the ridge line, a rotation matrix of the ridge line is solved by a least squares method, and a translation parameter of the ridge line is calculated by a distance field matching method;

[0044] When the translation parameter is less than the preset translation threshold, a plurality of pairs of ridge lines whose translation parameters are closest to the preset translation threshold are selected as the third matching results.

[0045] In some embodiments, according to the overall position deviation of the ridge points in the alignment area, or the local position deviation of the local position of the ridge points in the alignment area with the highest degree of alignment overlap when the overall position deviation of the ridge points in the alignment area exceeds a preset deviation range, a ridge point matching operation is performed on the two ridge line graphs to obtain a ridge point matching result, including:

[0046] Calculate the overall position deviation of the ridge points in the alignment area;

[0047] Determining whether the overall position deviation exceeds a preset deviation range;

[0048] If it does not exceed the limit, the ridge point matching result through which the ridge point matching passes is output;

[0049] If exceeded, when the local position deviation of the ridge point is less than the preset position deviation, a reference area including the ridge point is divided, and one of the two ridge maps is image-blocked to obtain a plurality of ridge map blocks;

[0050] Recursively inferring each ridge map block based on the reference area to obtain a target mapping transformation relationship; the target mapping transformation relationship is a rigid transformation relationship that minimizes the mapping position deviation between the ridge point in the reference area or in the ridge map block and the ridge point in the mapped ridge map after the ridge point is mapped to the other of the two ridge maps;

[0051] When the mapping position deviation is less than a preset mapping deviation threshold, a ridge point matching result through which the ridge point matching passes is output.

[0052] In some embodiments, the recursive derivation of each ridge line image block based on the reference area to obtain a target mapping transformation relationship includes:

[0053] Constructing a mapping-deviation problem; the mapping-deviation problem describes the subordinate relationship between the mapping position deviation and the mapping transformation relationship;

[0054] Calculate the rigid transformation relationship between the ridge points corresponding to the two ridge lines in the reference area to obtain the target mapping transformation relationship of the reference area;

[0055] Taking the reference area as the initial block, a block recursive operation is performed; the block recursive operation is to use the target mapping relationship of the adjacent blocks as the initial solution of the current ridge block about the mapping-deviation problem in a preset recursive direction, and solve the mapping transformation relationship that minimizes the mapping position deviation. As the operation of the target mapping transformation relationship, the adjacent blocks are the ridge blocks or the reference area adjacent to the current ridge block in the recursive direction.

[0056] The embodiment of the present application also provides a small-area fingerprint image matching device, comprising:

[0057] The first module is used to obtain at least two fingerprint images to be matched;

[0058] The second module is used to pre-process the fingerprint image to be matched to obtain a ridge line map containing ridge lines;

[0059] The third module is used to extract ridge point features from the ridge to obtain ridge point features;

[0060] The fourth module is used to extract the line features of the ridge line to obtain the line features of the ridge line;

[0061] A fifth module is used to perform a ridge alignment operation on the two ridge maps according to the ridge point features and the ridge body features to obtain an alignment area; the alignment area is an area where the two ridge maps partially overlap;

[0062] A sixth module is used to perform a ridge point matching operation on the two ridge line graphs according to the overall position deviation of the ridge line points in the alignment area, or the local position deviation of the local position of the ridge line points in the alignment area with the highest degree of alignment overlap when the overall position deviation of the ridge line points in the alignment area exceeds a preset deviation range, to obtain a ridge point matching result;

[0063] The seventh module is used to obtain the matching result of the two fingerprint images to be matched according to the ridge point matching result.

[0064] An embodiment of the present application further provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned small-area fingerprint image matching method when executing the computer program.

[0065] The embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned small-area fingerprint image matching method is implemented.

[0066] The beneficial effects of the present application are as follows: by extracting ridge point features and line body features from a ridge line graph after preprocessing of a fingerprint image to be matched, ridge point features and ridge line body features are obtained, a ridge line alignment operation is first performed on the two ridge line graphs while considering both ridge point features and ridge line body features, and an alignment area where the two ridge line graphs partially overlap is obtained, and then a ridge line point matching operation is performed on the two ridge line graphs while considering both the overall position deviation of the ridge line points in the alignment area and the local position deviation of the local position with the highest degree of alignment overlap of the ridge line points in the alignment area, and a ridge line point matching result is obtained, and based on the ridge line point matching result, a matching result of the two fingerprint images to be matched is obtained. Since the ridge point features and ridge body features are extracted at the same time, the feature elements required for fingerprint image matching are expanded, and the ridge alignment operation and the ridge point matching operation are performed successively. The ridge alignment operation takes into account the ridge point features and the ridge body features at the same time. The ridge point matching operation takes into account the overall position deviation of the ridge points in the alignment area and the local position deviation of the local position with the highest degree of ridge point alignment overlap in the alignment area. Multiple matches and comparisons can be performed step by step on the fingerprint images to be matched, overcoming the defects of the single fingerprint image matching mechanism and the single type of matching features, and improving the matching accuracy of small-area fingerprint images. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flow chart of a small-area fingerprint image matching method provided in an embodiment of the present application.

[0068] Figure 2 It is a flowchart of the specific method of step S103 provided in an embodiment of the present application.

[0069] Figure 3 It is a flowchart of the specific method of step S104 provided in an embodiment of the present application.

[0070] Figure 4 It is a flowchart of the specific method of step S105 provided in an embodiment of the present application.

[0071] Figure 5 It is a flowchart of the specific method of step S106 provided in an embodiment of the present application.

[0072] Figure 6 It is a structural schematic diagram of a small-area fingerprint image matching device provided in an embodiment of the present application.

[0073] Figure 7 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application.

[0074] Figure 8 It is a schematic diagram of the matching effect of two ridge line graphs provided in an embodiment of the present application.

[0075] Fig. 9It is a schematic diagram of the selection effect of the reference area provided in the embodiment of the present application.

[0076] Fig.10 It is a schematic diagram of the division effect of the ridge line block provided in an embodiment of the present application. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0078] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown can be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second" and the like in the specification, claims and drawings are used to distinguish similar objects, and are not used to describe a specific order or sequence.

[0079] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0080] See also Figure 1 , Figure 1 is a flow chart of a small area fingerprint image matching method provided in an embodiment of the present application. In some embodiments of the present application, Figure 1 The method may specifically include but is not limited to steps S101 to S107. Figure 1 These seven steps are introduced in detail.

[0081] Step S101, obtaining at least two fingerprint images to be matched.

[0082] Step S102, preprocessing the fingerprint image to be matched to obtain a ridge line map containing ridge lines.

[0083] Step S103, extracting ridge point features from the ridge to obtain ridge point features.

[0084] Step S104, extracting the line features of the ridges to obtain the line features of the ridges.

[0085] Step S105, performing a ridge alignment operation on the two ridge graphs according to the ridge point features and the ridge body features to obtain an alignment area.

[0086] The alignment area is the area where the two ridge graphs partially overlap.

[0087] Step S106, based on the overall position deviation of the ridge points in the alignment area, or the local position deviation of the local position with the highest degree of alignment overlap of the ridge points in the alignment area when the overall position deviation of the ridge points in the alignment area exceeds the preset deviation range, a ridge point matching operation is performed on the two ridge line graphs to obtain a ridge point matching result.

[0088] Step S107, obtaining the matching result of the two fingerprint images to be matched according to the ridge point matching result.

[0089] In step S101, the fingerprint can be collected by a fingerprint collection device to obtain a fingerprint image to be matched, and the collected fingerprint image to be matched is stored in a fingerprint database. By calling the fingerprint database and the currently collected fingerprint image to be matched, at least two fingerprint images to be matched can be obtained. For example, when using fingerprint attendance to punch in, multiple fingerprint images are stored on the attendance machine. When an employee punches in, he can enter his fingerprint on the attendance machine. The attendance machine collects the fingerprint to obtain a fingerprint image, and obtains the currently entered fingerprint image and at least one pre-stored fingerprint image to obtain at least two fingerprint images to be matched. It can be understood that the fingerprint image to be matched contains the fingerprint information obtained by collecting the fingerprint of a finger, and the fingerprint information in the two fingerprint images to be matched can be from the same finger or from different fingers.

[0090] In step S102, the fingerprint image to be matched is preprocessed, which may be noise filtering, directional field estimation, ridge enhancement, foreground and background segmentation, frequency field and mass field estimation, Gabor filter enhancement, binarization and thinning, so as to obtain a ridge map containing ridges. It can be understood that the raised lines that can be seen on the surface of the finger are generally called ridges, and the valleys between ridges are called valleys. Ridges are one of the important features of fingerprint images. The ridge map is a binary image, in which the fingerprint ridges have been converted into unit pixel widths, and the grayscale values ​​of the pixels on the ridge map are 0 or 1, and 1 represents a ridge point.

[0091] In step S103, the ridge point feature extraction of the ridge line can be performed by traversing each ridge point on the ridge line graph, classifying and marking the ridge point according to the position on the ridge line, and then dividing each ridge point according to the classification and marking result, so that a ridge point is divided into a ridge line, and the ridge point feature of the ridge point is obtained. For example, the ridge point can be classified into an online point, an endpoint, a bifurcation point and a boundary point according to the position on the ridge line, wherein the gray value of 1 is 1 and the number of neighbors is 1 and is not at the boundary of the ridge line graph. The number of neighbors of the gray value of 1 is 1 and is at the boundary of the ridge line graph is a boundary point, the number of neighbors of the gray value of 1 is 2. The number of neighbors of the gray value of 1 is greater than 2. A ridge line can be divided between two endpoints, between an endpoint and a boundary point, between an endpoint and a bifurcation point, and between a boundary point and a bifurcation point. The ridge points within this range are divided into this ridge line.

[0092] In step S104, the line features of the ridge are extracted, which may be represented by a ridge point set composed of ridge points, the ridge point set including the direction information and coordinate information of each ridge point, and the ridge line features are extracted by using the direction information and coordinate information of each ridge point included in the ridge to obtain the ridge line features. For example, the ridge line features may include line length features, curvature features, line width features and line type features, which may be obtained by accumulating and summing the coordinate information of each ridge point to obtain the line length feature of the ridge, which may be obtained by calculating the line curvature of the coordinate information and direction information of several adjacent ridge points to obtain the curvature feature of the ridge, which may be determined according to the direction information of the ridge point, and the ridge width in the tangent direction may be calculated to obtain the line width feature of the ridge, which may be determined according to the ridge point features of the first and last ridge points.

[0093] In step S105, the two ridgeline graphs are aligned, which can be based on the rigid transformation method. After traversing the ridgelines in the two ridgeline graphs, the ridgelines are prioritized according to the ridgeline point features and ridgeline body features, and the ridgelines with high feature stability and high discrimination power are set to have higher priorities. The alignment operation is performed on the ridgelines with the highest current priority first, and the alignment operation is performed on the ridgelines in order of priority. If there are a preset number of pairs of ridgelines that meet the corresponding alignment conditions during the alignment operation, the alignment operation is terminated. Otherwise, the alignment operation is continued until there are a preset number of pairs of ridgelines that meet the corresponding alignment conditions. After the ridgelines that meet the alignment conditions are aligned, the local overlap areas of the two ridgeline graphs are marked to obtain the alignment area. If there are no preset number of pairs of ridgelines that meet the corresponding alignment conditions when all ridgeline alignment is completed, the ridgelines of the two ridgeline graphs do not match, and the matching result of the mismatch of the fingerprint images to be matched is output. For example, it can be set that the ridge line whose end ridge point type is an endpoint or a bifurcation point has the highest priority, the ridge line with a maximum curvature greater than a preset maximum curvature threshold other than the highest priority has the second highest priority, and the ridge line other than the second highest priority has the lowest priority. If both ridge line graphs have a ridge line with the highest priority, the ridge line with the highest priority is aligned first. If there are a preset number of ridge lines that meet the corresponding alignment conditions, the alignment is terminated and it is determined that the ridge lines of the two ridge line graphs are alignable. If there is no ridge line with the highest priority or the ridge line with the highest priority cannot be aligned, if both ridge line graphs do not have a ridge line with the second highest priority , the alignment operation is performed on the ridges with the second highest priority first. If there are a preset number of pairs of ridges that meet the corresponding alignment conditions, the alignment is terminated and it is determined that the ridges of the two ridge line graphs can be aligned. When there are only ridges with the lowest priority or the ridges with the highest priority and the second highest priority cannot be aligned, the alignment operation is performed on the ridges with the lowest priority. If there are a preset number of pairs of ridges that meet the corresponding alignment conditions, the alignment is terminated and it is determined that the ridges of the two ridge line graphs can be aligned. If at the end of the alignment of all ridge lines, there is no preset number of pairs of ridges that meet the corresponding alignment conditions, the ridges of the two ridge line graphs do not match, and the matching result that the fingerprint images to be matched do not match is output.

[0094] In step S106, a ridge point matching operation is performed on the two ridge line graphs, which can be performed by simultaneously considering the overall position deviation of the ridge line points in the alignment area and the local position deviation of the local position with the highest degree of ridge line point alignment overlap in the alignment area, so as to perform a ridge point matching operation on the two ridge line graphs. When the overall position deviation does not exceed the preset deviation range, a ridge point matching result through which the ridge line points of the two ridge line graphs are matched is obtained. When the overall position deviation exceeds the preset deviation range, a ridge point matching operation is performed on the two ridge line graphs according to the local position deviation of the local position with the highest degree of ridge line point alignment overlap in the alignment area. If the alignment area The local position deviation of the local position with the highest degree of ridge point alignment overlap in the domain exceeds the preset deviation range, and the ridge point matching result of the two ridge line graphs is obtained. If the local position deviation of the local position with the highest degree of ridge point alignment overlap in the alignment area does not exceed the preset deviation range, the ridge points in the two ridge line graphs are rigidly transformed based on the local position with the highest degree of alignment overlap to determine the overall position deviation of the ridge points in the two ridge line graphs. According to the overall position deviation of the ridge points in the two ridge line graphs, it is determined whether the ridge points in the two ridge line graphs match to obtain the corresponding ridge point matching result.

[0095] In step S107, when the ridge point matching result of the two ridge line graphs is obtained, the matching result of the two fingerprint images to be matched is obtained; when the ridge point matching result of the two ridge line graphs is obtained, the matching result of the two fingerprint images to be matched is obtained.

[0096] In steps S101 to S107 of the embodiment of the present application, ridge point features and line body features are extracted from the ridge line graph after preprocessing of the fingerprint image to be matched, so as to obtain ridge point features and ridge line body features. First, a ridge alignment operation is performed on the two ridge line graphs while considering both the ridge point features and the ridge line body features to obtain an alignment area where the two ridge line graphs partially overlap. Then, a ridge point matching operation is performed on the two ridge line graphs while considering both the overall position deviation of the ridge points in the alignment area and the local position deviation of the local position with the highest degree of ridge point alignment overlap in the alignment area to obtain a ridge point matching result. Based on the ridge point matching result, a matching result of the two fingerprint images to be matched is obtained. Since the ridge point features and ridge body features are extracted at the same time, the feature elements required for fingerprint image matching are expanded, and the ridge alignment operation and the ridge point matching operation are performed successively. The ridge alignment operation takes into account the ridge point features and the ridge body features at the same time. The ridge point matching operation takes into account the overall position deviation of the ridge points in the alignment area and the local position deviation of the local position with the highest degree of ridge point alignment overlap in the alignment area. Multiple matches and comparisons can be performed step by step on the fingerprint images to be matched, overcoming the defects of the single fingerprint image matching mechanism and the single type of matching features, and improving the matching accuracy of small-area fingerprint images.

[0097] See also Figure 2 In a specific embodiment, step S103 may specifically include but is not limited to steps S201 to S203. Figure 2 These three steps are introduced in detail.

[0098] Step S201, extracting ridge points in the ridge map.

[0099] Step S202, classifying the ridge points according to the neighborhood colors of the ridge points and the positions in the ridge map, to obtain ridge point category features.

[0100] Step S203, tracking the ridge line according to the ridge point category feature to obtain a tracking result, and correlating the ridge line and the ridge point according to the tracking result to obtain a ridge point correlation feature.

[0101] In step S201, ridge points in the ridge map are extracted by performing interval sampling along the length direction of the ridge according to the grayscale values ​​of the pixels in the ridge map, and performing ridge point sampling once between intervals along the length direction of the ridge to obtain multiple required ridge points.

[0102] In step S202, the ridge points are classified according to the neighborhood colors of the ridge points and the positions in the ridge map. The eight-neighborhood algorithm may be used to determine the grayscale values ​​of the eight neighborhoods of the ridge points, and the ridge points are classified according to the number of grayscale values ​​of 1 in the eight neighborhoods of the ridge points to obtain the ridge point category features. Specifically, the ridge point category features include online points, endpoints, bifurcation points, and boundary points, wherein the points with a grayscale value of 1 and a number of neighbors that are not at the boundary of the ridge map are endpoints, the points with a grayscale value of 1 and a number of neighbors that are at the boundary of the ridge map are boundary points, the points with a grayscale value of 1 and a number of neighbors that are 2 are online points, and the points with a grayscale value of 1 and a number of neighbors that are greater than 2 are bifurcation points.

[0103] In step S203, a ridge tracking algorithm may be used to extract each ridge in the ridge map, and ridge tracking may be performed according to the ridge point category features of each ridge point. The divided ridges and the ridge points on the ridges may be associated to obtain ridge point association features. Specifically, ridge tracking is to start tracking from a ridge point in the ridge map that has not been tracked and whose ridge point category feature is an endpoint or a boundary point. If there are no endpoints and boundary points among the ridge points that have not been tracked, ridge tracking is randomly started from a bifurcation point, and the adjacent ridge points of the current tracking point are identified to identify the number of untracked ridge points in the eight neighborhoods of the current tracking point. If the number of untracked ridge points is 1, the current tracking point is marked as a tracked point, and the tracking position is moved to the adjacent ridge point. If the number of untracked ridge points is greater than 1, the current tracking point is not marked as a tracked point, and the tracking position is moved to one of the untracked adjacent ridge points of the current tracking point. If the number of untracked ridge points is 0, the current tracking point is marked as a ridge termination point, the current ridge tracking is ended, the ridge points included in the currently tracked ridge are determined, and the next ridge tracking is started when there are still untracked ridge points, until all ridge points are tracked.

[0104] See also Figure 3 In a specific embodiment, the ridge line features include line length features, curvature features, line width features and line type features. Step S104 may specifically include but is not limited to steps S301 to S306. Figure 3 These six steps are introduced in detail.

[0105] Step S301, perform Freeman chain code encoding on the ridge to obtain a ridge chain code representation.

[0106] Step S302, performing mapping accumulation operation on the ridge chain code representation to obtain the line length feature.

[0107] Step S303, calculating the curvature of the ridge line according to the relative positions of the ridge line points on the ridge line to obtain the curvature feature.

[0108] Step S304, taking the sampling points on the ridge line as a reference, the ridge line is subjected to line width detection in a direction perpendicular to the ridge line to obtain line width features.

[0109] Step S305, classifying the ridges according to the target ridge point features to obtain line features.

[0110] The target ridge point features are ridge point features of ridge points located at the beginning and end positions of the ridge line.

[0111] Step S306, when the linear feature characterizes the ridge as the target ridge category, the ridge is sampled at intervals, and a corresponding ridge descriptor is generated according to the interval sampling result.

[0112] In step S301, the ridges are encoded by Freeman chain codes. After tracking and extracting several ridges, the ridge points of each ridge are encoded by using eight-directional Freeman chain codes, and the ridge points are sorted according to their positions on the ridges. The position coordinates of the starting ridge points and the direction information of each ridge point are recorded to obtain the ridge chain code representation. For example, S can be used to represent the position coordinates of the starting ridge point, and for the i-th ridge point on the ridge, it can be used To represent the ridge point encoding representation of the ridge point, the ridge point encoding representation Indicates the direction from the i-th ridge point to the i+1-th ridge point. , then for a specific ridge line, its ridge line chain code representation can be expressed as , n is the number of ridge points in the ridge.

[0113] In step S302, the calculation formula for mapping and accumulating the ridge chain code representation is:

[0114] ,

[0115] ,

[0116] in, is the line length feature, is the chain code mapping result of the i-th ridge point.

[0117] In step S303, the curvature of the ridge line is calculated by traversing the ridge line points at non-starting positions and non-ending positions on the ridge line, and combining the curvature calculation of two adjacent ridge line points or ridge line points separated by a certain length and the position coordinates of the ridge line point to obtain the curvature feature. Specifically, for the i-th ridge line point in the ridge line For example, the ridge point and two ridge points with a spacing of m (m is a positive integer, such as 5) from the ridge point are combined. and The curvature is calculated at the position coordinates, where The position coordinates are , The position coordinates are , The position coordinates are , then the calculation formula for the curvature of the ridge line is:

[0118] ,

[0119] ,

[0120] in, for , and The area of ​​the triangle formed by connecting the two lines, is the curvature characteristic of the ridge line at the i-th ridge point.

[0121] In step S304, the ridge line width is detected, and the ridge line direction can be obtained through the direction field. In the ridge line diagram, the ridge line points are checked along the direction perpendicular to the ridge line direction. When a non-ridge line pixel point is encountered, the check stops. The obtained line segment length is the ridge line width of the point, and the local line width feature of the ridge line is obtained. After traversing each ridge line point on the ridge line or a number of ridge line points pre-selected as samples, the overall line width feature of the ridge line is obtained.

[0122] In step S305, the ridges are classified according to the target ridge point features, and the ridges can be classified according to the ridge point features of the first and last ridge points to determine the linear features of the ridges. Specifically, when the ridge point features of the ridge points located at the beginning and end of the ridge are the target ridge point features, the linear features of the ridge are the target ridge features, and when the ridge point features of the ridge points located at the beginning and end of the ridge are non-target ridge point features, the linear features of the ridge are non-target ridge features. For example, when the endpoint and the bifurcation point are the target ridge point features, and the starting ridge point or the ending ridge point of the ridge is the endpoint or the bifurcation point, the linear features of the ridge are the target ridge features, otherwise, the linear features of the ridge are non-target ridge features.

[0123] In step S306, the ridges are sampled at intervals, and a corresponding ridge descriptor is generated based on the interval sampling results. The ridges may be sampled and marked at certain intervals starting from the starting ridge point of the ridge, and a circular area is constructed with each sampling point as the center and a preset parameter as the radius. The direction obtained by the sampling point in the direction field is taken as the horizontal direction of the area. The sampling point descriptor is obtained by performing specific calculations and statistics on the grayscale values ​​in this circular area. The circular area constructed by all sampling points is called the support area of ​​the ridge descriptor. Then, in the circular area constructed by each sampling point, the ridge descriptor is generated by performing histogram statistics on the gradient. Due to the selection rules of the horizontal direction of the region and the characteristics of the fingerprint ridges, the gradient near the vertical direction is more significant. The gradient histogram statistics of this embodiment are performed near the horizontal direction. The histogram has 8 columns. The gradient angle range of each column is [0°, 30°], [30°, 60°], [120°, 150°], [150°, 180°], [180°, 210°], [210°, 240°], [300°, 330°] and [330°, 360°]. Each sampling point is counted to obtain an 8-dimensional statistic. , for each dimension component , calculate its sample mean , and then use the mean to perform binary quantization on each dimension to obtain the sampling point descriptor ,in for:

[0124] ,

[0125] Therefore, each sampling point can obtain an 8-bit binary sampling point descriptor, and these sampling point descriptors are connected in series to obtain the ridge descriptor of the ridge.

[0126] See also Figure 4 In a specific embodiment, the ridge line features include line length features, curvature features, line width features and line type features. Step S105 may specifically include but is not limited to steps S401 to S404. Figure 4 These four steps are introduced in detail.

[0127] Step S401, when the ridges meet the first alignment condition, similarity matching and rigid transformation matching are performed on the ridges according to the line length feature, and several pairs of ridges with the highest matching degree are determined to obtain a first matching result.

[0128] Step S402, when the ridge line meets the second alignment condition, the inflection point of the ridge line is located according to the curvature feature, and the ridge line is matched by line coincidence and rigid transformation using the located inflection point to determine several pairs of ridge lines with the highest matching degree and obtain a second matching result.

[0129] Step S403, when the ridge lines do not meet the first alignment condition and the second alignment condition, the ridge lines are rigidly transformed and matched and the transformation angles are estimated according to the curvature characteristics and line width characteristics, and the ridge lines are rotated to the estimated transformation angles and then line coincidence matching is performed to determine the pairs of ridge lines with the highest matching degree and obtain a third matching result.

[0130] Step S404: align the two ridge line graphs according to the first matching result, the second matching result or the third matching result to form an alignment area.

[0131] Among them, the first alignment condition is that the starting ridge point and / or the ending ridge point of the ridge line are formed by non-linear interruption or by linear bifurcation, and the second alignment condition is that the maximum curvature of the ridge line is greater than a preset maximum curvature threshold.

[0132] In step S401, for ridges that meet the first alignment condition, the starting ridge point and / or the ending ridge point are formed by non-linear interruption or by linear bifurcation, that is, the starting ridge point and / or the ending ridge point of the ridge are endpoints or bifurcation points. When a ridge alignment operation is performed on two such ridges, the line length deviation between the two ridges is calculated based on the line length characteristics of the two ridges. When the line length deviation meets the preset line length deviation condition, the ridge descriptors of the two ridges are similarly matched. When the result of the ridge descriptor similarity matching meets the preset descriptor similarity condition, the two ridges are rigidly transformed and matched, and the rotation matrix and translation parameters required for the alignment of the two ridges are solved. After traversing multiple pairs of ridges, the optimal sets of rotation matrices and translation parameters are selected to obtain the first matching result.

[0133] In some embodiments, step S401 may include the following steps:

[0134] According to the line length characteristics, the line length deviation between the ridge lines is calculated;

[0135] When the line length deviation is less than a preset line length deviation threshold, similarity matching is performed on the ridge descriptors of the ridges to obtain the descriptor similarity;

[0136] When the similarity of the descriptor is greater than the preset similarity threshold, the ridge line is rigidly transformed and matched, and the rotation matrix and translation parameters of the ridge line are solved by the least squares method;

[0137] When the rotation matrix is ​​smaller than the preset rotation threshold and the translation parameter is smaller than the preset translation threshold, several pairs of ridge lines whose rotation matrix is ​​closest to the preset rotation threshold and whose translation parameter is closest to the preset translation threshold are selected as the first matching results.

[0138] Specifically, for ridges that meet the first alignment condition, when the line length deviation of the two ridges is less than the preset line length deviation threshold, the descriptor similarity of the ridge descriptor can be used to measure the ridge similarity to determine whether the two ridges match. When the descriptor similarity is greater than the preset similarity threshold, the two ridges are preliminarily matched. and The calculation formula of the descriptor similarity is:

[0139] ,

[0140] in, is the descriptor similarity, Ridge The ridge descriptor of Ridge The ridge descriptor of Ridge The line length, Ridge The line length, is the exclusive OR operator, Ridge The ridge point descriptor of the i-th ridge point, Ridge The ridge point descriptor of the i-th ridge point.

[0141] In some embodiments, the ridge points at the starting and ending positions of the ridge are endpoints and / or bifurcation points. In some scenarios that require 360-degree recognition, it is necessary to calculate the ridge descriptors of this type of ridge in two directions, namely, the ridge descriptor obtained by sampling from the initial ridge point and the ridge descriptor obtained by sampling in the reverse direction from the ending ridge point. When calculating the ridge similarity, the ridge similarity needs to be calculated twice.

[0142] When the descriptor similarity is greater than the preset similarity threshold, a rigid transformation relationship between the two ridges is constructed and calculated. The expression of the rigid transformation relationship is:

[0143] ,

[0144] in, and are the ridge point sets of the two ridges, R is the rotation matrix in the rigid transformation, and T is the translation parameter in the rigid transformation.

[0145] Combining the calculation formula of the descriptor similarity and the expression of the rigid transformation relationship, an overdetermined set of equations about the rotation matrix and translation parameters can be obtained. Solving the least squares solution of the set of equations can obtain the rigid transformation relationship of this pair of matching ridges, that is, the pair of ridges is matched. If the two ridge graphs only match no more than a preset number of pairs of ridges that meet the first alignment condition, the calculated rotation matrix and translation parameters are used as the first matching result. If no more than a preset number of pairs of ridges that meet the first alignment condition are matched, check whether the rigid transformation relationship obtained by matching each pair of ridges is compatible. The compatibility is determined by calculating the deviation of the rotation matrix and translation parameters of the rigid transformation relationship. If the deviation of the rotation matrix and translation parameters exceeds a certain preset threshold, it is determined that the rigid transformation relationship is incompatible, otherwise it is compatible. Find the rigid transformation relationship with the most compatible matching pairs, and use the corresponding rotation matrix and translation parameters as the first matching result. If no ridge line that meets the first alignment condition is matched, the two ridge line graphs are checked to determine whether the two ridge line graphs contain a preset number of ridge lines that meet the first alignment condition and whether the preset image area contains ridge lines that meet the first alignment condition. For example, it is determined whether the two ridge line graphs contain at least four ridge lines that meet the first alignment condition and whether each quadrant area of ​​the four-quadrant coordinate system constructed with the center of the ridge line graph contains at least one ridge line that meets the first alignment condition. If at least one ridge line graph is yes, the ridge lines of the two ridge line graphs do not match, and a matching result that the fingerprint images to be matched do not match can be obtained.

[0146] In step S402, for ridges that meet the second alignment condition, the maximum curvature of the ridges is used to match the ridges. All the inflection points of the ridges are calculated and the ridge matching is approximated by inflection point matching to perform line coincidence matching on the ridges, thereby quickly calculating the overlapping parts of the ridges. After the overlapping parts of the ridges are matched, the corresponding inflection points are in a corresponding relationship, which is converted into a rigid transformation matching of the ridges, and the rotation matrix and translation parameters required for the alignment of the two ridges are solved. After traversing multiple pairs of ridges, the optimal sets of rotation matrices and translation parameters are selected to obtain the second matching result.

[0147] In some embodiments, step S402 may include the following steps:

[0148] According to the curvature characteristics, the maximum curvature of the ridge line is determined;

[0149] When the maximum curvature is greater than a preset maximum curvature threshold, calculating the maximum curvature deviation between ridge lines;

[0150] When the maximum curvature deviation is less than the preset curvature deviation threshold, the chain code derivative integration method is used to locate the inflection point of the ridge line, and the inflection point is located;

[0151] According to the number and relative angle of inflection points, the degree of line coincidence matching of the ridge line is calculated;

[0152] When the line weight coincidence matching degree is less than a preset line weight coincidence threshold, a rigid transformation matching is performed on the ridge lines to determine several pairs of ridge lines with the highest matching degree, and obtain a second matching result.

[0153] Specifically, for two ridges that meet the second alignment condition and , ridge line and The maximum curvatures are and , only when the maximum curvature deviation When the curvature is less than a certain preset maximum curvature threshold, the ridge and Match. Determine the ridge line and To find out whether the ridge line matches, we first use the chain code derivative integral method to locate the inflection point on the ridge line. Here, the chain code derivative The definition is as follows:

[0154] ,

[0155] ,

[0156] in, is the differential code, is the chain code derivative, and All are ridge point codes. represents the modulo operation. Starting from the second ridge point of the ridge line, the chain code derivative integral is calculated. Then, from the second ridge point to the The chain code derivative integral of the ridge points for:

[0157] ,

[0158] when When The ridge point is the inflection point. When the inflection point is extracted, the initial position of the cumulative summation of the chain code derivative becomes the inflection point position plus 1.

[0159] Assumptions , is the position of the first two inflection points of the ridge line, then:

[0160] ,

[0161] ,

[0162] After calculating all the inflection points of the ridges that meet the second alignment condition, the ridge matching can be approximated by inflection point matching. Assuming , , ,……, For all the inflection points on the ridge line obtained in sequence, the inflection point length sequence of the ridge line is defined as and the inflection point angle sequence as follows:

[0163] ,

[0164] ,

[0165] in, is the length of the interval between the i-th inflection points, is the relative angle of the i-th inflection point. From this, the ridge line can be calculated and The overlap segment of is the length of the overlapped segment sequence that needs to be checked, , Represents their respective starting positions, then the difference calculation formula of the overlapped segment is:

[0166] ,

[0167] Among them, T is the line coincidence matching degree of the coincident segment, is the weight parameter, is the relative angle of the i-th inflection point of ridge line a, is the relative angle of the i-th inflection point of the ridge line b, is the interval length of the i-th inflection point of ridge line a, is the length of the interval between the i-th inflection points of ridge line b. The preset line coincidence threshold is T th , when T<T th When and The coincident segments of the and The overlapping segments of and The corresponding inflection points are matched by rigid transformation matching of the ridge line, which is converted into the least square solution problem of the overdetermined equations of the rotation matrix and translation parameters of the inflection line, so as to find the ridge line. and According to the rigid transformation relationship between the two ridges, after traversing multiple pairs of ridges, the optimal groups of rotation matrices and translation parameters are selected to obtain the second matching result. If no ridge line that meets the second alignment condition is matched, the two ridge line graphs are checked to determine whether the two ridge line graphs contain a preset number of ridge lines that meet the second alignment condition. For example, it is determined whether the proportion of the number of ridge lines that meet the second alignment condition in the two ridge line graphs exceeds 70%. If at least one ridge line graph is yes, the ridges of the two ridge line graphs do not match, and a matching result that the fingerprint images to be matched do not match can be obtained.

[0168] In step S403, for ridges that do not meet both the first alignment condition and the second alignment condition, the ridges with large differences in curvature characteristics and line width characteristics are first filtered out based on the curvature characteristics and line width characteristics, and then the ridges are rigidly transformed and matched and the transformation angles are estimated. First, the angle for the required rigid transformation matching is estimated, and then after the corresponding rotation angle is estimated, the ridge is rotated to the estimated transformation angle, and the translation parameters that need to be verified are calculated by enumeration. The calculated translation parameters are used to perform line coincidence matching on the ridges to determine several pairs of ridges with the highest matching degree, and a third matching result is obtained.

[0169] In some embodiments, step S403 may include the following steps:

[0170] According to the curvature characteristics and line width characteristics, the average curvature and average line width of the ridge line are calculated;

[0171] When the average curvature is greater than a preset average curvature threshold and the average line width is greater than a preset average line width threshold, a rigid transformation matching is performed on the ridge line, the rotation matrix of the ridge line is solved by the least squares method, and the translation parameters of the ridge line are calculated by the distance field matching method;

[0172] When the translation parameter is less than the preset translation threshold, several pairs of ridge lines whose translation parameters are closest to the preset translation threshold are selected as the third matching results.

[0173] According to the curvature characteristics and line width characteristics, the average curvature and average line width of the ridges are calculated, and the calculated average curvature and average line width are used to filter out the matches between obviously dissimilar ridges. Then, the angle of the required rigid transformation is estimated first. This is because the general ridges are relatively flat at present, and each ridge can be basically regarded as a straight line. For each possible matching ridge matching pair, its corresponding rotation angle is calculated, and the rotation angle between the two currently matched small-area fingerprint images can be solved by statistically analyzing the rotation angle histogram. After estimating the rotation angle, considering the condition that the overlapping area needs to reach the preset threshold, the enumeration method is used to calculate the translation parameters that need to be verified. The ridge pre-pairing method can filter out a large amount of parameter space that does not need to be searched, such as Figure 8 As shown, in a specific embodiment, after the rotation angle is estimated, both ridgeline graphs are rotated until the ridgeline is in a horizontal state as a whole, so that the corresponding number of ridgelines in the two ridgeline graphs match each other, and the ridgeline in the middle position of one ridgeline graph can only match the ridgeline in the middle position of the other ridgeline graph, for example, the first ridgeline in the ridgeline graph on the lower right side can only match the ridgelines 1 to 11 in the ridgeline graph on the upper left side, otherwise it will be impossible to identify the true or false due to too little information, thereby determining the value of the translation parameter in the vertical direction. After determining the rotation angle of the rigid deformation and the vertical displacement after the correction, the displacement in the horizontal direction is quickly solved by matching based on the distance field. First, the distance field of one ridgeline graph is calculated using the chamfer distance transformation, and then for each horizontal displacement that needs to be verified, the overall distance deviation obtained in the distance field of the ridgeline graph after acting on the other ridgeline graph is calculated, and the horizontal displacement corresponding to the minimum distance deviation is taken as the optimal solution. If the minimum distance deviation is greater than the preset translation threshold, a matching result that the fingerprint image to be matched does not match can be obtained.

[0174] In step S404, when the first matching result, the second matching result or the third matching result is obtained, the two ridgeline graphs are aligned so that the ridges in the two ridgeline graphs that meet the first matching result, the second matching result or the third matching result are aligned to form an alignment area.

[0175] See also Figure 5 In a specific embodiment, step S106 may specifically include but is not limited to steps S501 to S506. Figure 5 These six steps are introduced in detail.

[0176] Step S501, calculating the overall position deviation of the ridge points in the alignment area.

[0177] Step S502, determine whether the overall position deviation exceeds a preset deviation range. If not, execute step S503; if exceeded, execute step S504.

[0178] Step S503, outputting the ridge point matching result after the ridge point matching is completed.

[0179] Step S504, when the local position deviation of the ridge point is less than the preset position deviation, a reference area including the ridge point is divided, and one of the two ridge images is divided into blocks to obtain a plurality of ridge image blocks.

[0180] Step S505, recursively deducing each ridge line image block based on the reference area to obtain a target mapping transformation relationship.

[0181] The target mapping transformation relationship is a rigid transformation relationship that minimizes the deviation of the mapping position between the ridge points in the reference area or in the ridge map block and the ridge points in the mapped ridge map after the ridge points are mapped to the other of the two ridge maps.

[0182] Step S506: when the mapping position deviation is less than a preset mapping deviation threshold, output the ridge point matching result through which the ridge point matching passes.

[0183] In step S501 to step S506, the distance field measurement method is used to calculate the overall position deviation of the ridge points in the alignment area. When the overall position deviation of the ridge points in the alignment area exceeds the preset deviation range, the ridge point matching result in which the ridge point matching fails is output. On the basis that the overall position deviation of the ridge points in the alignment area does not exceed the preset deviation range, the reference area (such as the local position deviation of the ridge points in the alignment area) where the local position deviation of the ridge points is less than the preset position deviation is divided according to the local position deviation of the ridge points in the alignment area. Fig. 9 The area in the box shown is the reference area). The local position deviation of the ridge points in the reference area is smaller than the local position deviation of the ridge points outside the reference area. Based on the reference area, one of the two ridge maps is divided into blocks. Fig.10 The multiple ridge line blocks shown are then based on the rigid transformation relationship between the two ridge line maps in the reference area, and the target mapping transformation relationship of mapping the ridge line points in each ridge line block to the other of the two ridge line maps is recursively deduced. After mapping the ridge line points in the ridge line block to the other of the two ridge line maps according to the target mapping transformation relationship, the mapping position deviation between the ridge line points mapped from the ridge line block and the ridge line points in the mapped ridge line map is calculated. When the mapping position deviation is less than a preset mapping deviation threshold, the ridge line point matching result of the ridge line point matching is output, otherwise, the ridge line point matching result of the ridge line point matching is output.

[0184] In some embodiments, step S505 may include the following steps:

[0185] Construct mapping-bias problem;

[0186] The mapping-deviation problem describes the subordination between the mapping position deviation and the mapping transformation relation;

[0187] Calculate the rigid transformation relationship between the ridge points corresponding to the two ridge graphs in the reference area to obtain the target mapping transformation relationship of the reference area;

[0188] Taking the reference area as the initial block, perform the tile recursive operation;

[0189] The tile recursive operation is to use the target mapping relationship of adjacent tiles as the initial solution to the mapping-bias problem of the current ridge tile according to the preset recursive direction, and solve the mapping transformation relationship that minimizes the mapping position deviation. As the operation of the target mapping transformation relationship, the adjacent tiles are the ridge tiles or reference areas adjacent to the current ridge tile in the recursive direction.

[0190] For example, for ridgeline graphs A and B, ridgeline graph B is divided into multiple ridgeline graph blocks, and each ridgeline point in ridgeline graph B belongs only to the ridgeline graph block that contains it.

[0191] Definition Ridge tiles The ridge point set in is , each ridge line block has a transformation relationship that maps the ridge line point set in the ridge line block to the coordinate system of the ridge line map A. Let Ridge tiles The transformation relationship is , for the Ridge tiles Any ridge point in , the transformation relationship of the point mapped to the coordinate system of the ridge line graph A is .

[0192] For the pixel points mapped to the coordinate system of the ridge map A , the error of the pixel is defined as:

[0193] ,

[0194] in, is the set of ridge points of ridge map A, v is the pixel point after the ridge point is mapped from ridge map B to ridge map A, Represents pixel and The Euclidean distance of .

[0195] For the first Ridge tiles , define an error for:

[0196] ,

[0197] when The smaller the Ridge tiles The higher the degree of match.

[0198] For Ridge tiles , construct the mapping-deviation problem , where the mapping-bias problem describes the subordinate relationship between the mapping position deviation and the mapping transformation relationship, and is the process of finding the mapping transformation relationship of the ridge point when the mapping position deviation is minimized. Specifically:

[0199] Min: ,

[0200] Subject to:

[0201] ,

[0202] in, It is the ridge point The coordinates in the coordinate system of the ridge graph B are expressed as follows: is a rigid transformation The parameter variables of the mapping-deviation problem are given by the initial value solution ( ), a more accurate solution is obtained by iterating N times through the steepest descent method ( ).

[0203] For two adjacent ridge blocks in ridge map B (adjacent left and right or adjacent up and down), and , defined from arrive Recursively, The transformation relationship As middle Initial solution to the problem, solving the mapping-deviation problem , the optimal solution obtained , as the target mapping transformation relationship, and finally obtain the target mapping transformation relationship of all ridge blocks.

[0204] In a specific embodiment, a tile recursive operation is performed with the reference area as the initial block, which may be to use the reference area as the initial block and recurse to the right until it cannot be recursed to the right, use the reference area as the initial block and recurse to the left until it cannot be recursed to the left, use each ridge line tile of the row where the reference area is located as the initial block and recurse upward until it cannot be recursed upward, and use each ridge line tile of the row where the reference area is located as the initial block and recurse downward until it cannot be recursed downward.

[0205] See also Figure 6 The embodiment of the present application also provides a small-area fingerprint image matching device, which can implement the above-mentioned small-area fingerprint image matching method, and the device includes:

[0206] The first module 601 is used to obtain at least two fingerprint images to be matched;

[0207] The second module 602 is used to pre-process the fingerprint image to be matched to obtain a ridge line map containing ridge lines;

[0208] The third module 603 is used to extract ridge point features from the ridges to obtain ridge point features;

[0209] The fourth module 604 is used to extract the line features of the ridges to obtain the line features of the ridges;

[0210] The fifth module 605 is used to perform a ridge alignment operation on the two ridgeline graphs according to the ridgeline point features and the ridgeline body features to obtain an alignment area; the alignment area is an area where the two ridgeline graphs partially overlap;

[0211] The sixth module 606 is used to perform a ridge point matching operation on the two ridge line graphs according to the overall position deviation of the ridge line points in the alignment area, or the local position deviation of the local position with the highest degree of alignment overlap of the ridge line points in the alignment area when the overall position deviation of the ridge line points in the alignment area exceeds a preset deviation range, to obtain a ridge point matching result;

[0212] The seventh module 607 is used to obtain the matching result of the two fingerprint images to be matched according to the ridge point matching result.

[0213] The specific implementation of the small-area fingerprint image matching device is substantially the same as the specific implementation of the small-area fingerprint image matching method described above, and will not be described in detail herein.

[0214] Figure 7 It is a block diagram of an electronic device according to an exemplary embodiment.

[0215] Refer to the following Figure 7 An electronic device 700 according to this embodiment of the present disclosure is described. Figure 7 The electronic device 700 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0216] like Figure 7 As shown, the electronic device 700 is in the form of a general computing device. The components of the electronic device 700 may include, but are not limited to: at least one processing unit 710, at least one storage unit 720, a bus 730 connecting different system components (including the storage unit 720 and the processing unit 710), a display unit 740, etc.

[0217] The storage unit stores program codes, which can be executed by the processing unit 710, so that the processing unit 710 executes the steps according to various exemplary embodiments of the present disclosure described in the above small-area fingerprint image matching method section of this specification.

[0218] The storage unit 720 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 7201 and / or a cache storage unit 7202 , and may further include a read-only storage unit (ROM) 7203 .

[0219] The storage unit 720 may also include a program / utility 7204 having a set (at least one) of program modules 7205, such program modules 7205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0220] Bus 730 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0221] The electronic device 700 may also communicate with one or more external devices 700' (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 700, and / or communicate with any device that enables the electronic device 700 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 750. Furthermore, the electronic device 700 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 760. The network adapter 760 may communicate with other modules of the electronic device 700 via a bus 730. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0222] The embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned small-area fingerprint image matching method is implemented.

[0223] The small-area fingerprint image matching method, apparatus, device and storage medium provided in the embodiments of the present application obtain ridge point features and ridge body features by extracting ridge point features and line body features from a ridge line graph after preprocessing of the fingerprint image to be matched, firstly perform a ridge line alignment operation on the two ridge line graphs that simultaneously considers the ridge point features and the ridge body features, and obtains an alignment area where the two ridge line graphs partially overlap, then perform a ridge line point matching operation on the two ridge line graphs that simultaneously considers the overall position deviation of the ridge line points in the alignment area and the local position deviation of the local position with the highest degree of ridge line point alignment overlap in the alignment area, and obtains a ridge line point matching result, and obtains a matching result of the two fingerprint images to be matched based on the ridge line point matching result. Since the ridge point features and ridge body features are extracted at the same time, the feature elements required for fingerprint image matching are expanded, and the ridge alignment operation and the ridge point matching operation are performed successively. The ridge alignment operation takes into account the ridge point features and the ridge body features at the same time. The ridge point matching operation takes into account the overall position deviation of the ridge points in the alignment area and the local position deviation of the local position with the highest degree of ridge point alignment overlap in the alignment area. Multiple matches and comparisons can be performed step by step on the fingerprint images to be matched, overcoming the defects of the single fingerprint image matching mechanism and the single type of matching features, and improving the matching accuracy of small-area fingerprint images.

[0224] Through the description of the above implementation, it is easy for technicians in this field to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the above method according to the implementation of the present disclosure.

[0225] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0226] Computer readable storage media may include data signals propagated in baseband or as part of a carrier wave, wherein readable program codes are carried. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program codes contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0227] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0228] Those skilled in the art will appreciate that the above modules can be distributed in the device according to the description of the embodiment, or can be changed to one or more devices different from the embodiment. The modules of the above embodiments can be combined into one module, or further divided into multiple sub-modules.

[0229] Through the description of the above embodiments, it is easy for technicians in this field to understand that the example embodiments described here can be implemented by software, or by combining software with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0230] The exemplary embodiments of the present disclosure are specifically shown and described above. It should be understood that the present disclosure is not limited to the detailed structures, configurations or implementations described herein; on the contrary, the present disclosure is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended claims.

Claims

1. A small area fingerprint image matching method, characterized in that: include: Obtain at least two fingerprint images to be matched; Preprocessing the fingerprint image to be matched to obtain a ridge line map containing ridge lines; Extracting ridge point features from the ridge to obtain ridge point features; Extracting line features of the ridge line to obtain ridge line features; According to the ridge point features and the ridge body features, a ridge alignment operation is performed on the two ridge graphs to obtain an alignment area; the alignment area is an area where the two ridge graphs partially overlap; According to the overall position deviation of the ridge points in the alignment area, or the local position deviation of the local position of the ridge points in the alignment area with the highest degree of alignment overlap when the overall position deviation of the ridge points in the alignment area exceeds a preset deviation range, a ridge point matching operation is performed on the two ridge line graphs to obtain a ridge point matching result; According to the ridge point matching result, obtaining the matching result of the two fingerprint images to be matched; The ridge line body features include line length features, curvature features, line width features and line type features. The ridge line alignment operation is performed on the two ridge line graphs according to the ridge line point features and the ridge line body features to obtain an alignment area, including: When the ridge line meets the first alignment condition, similarity matching and rigid transformation matching are performed on the ridge line according to the line length feature, and several pairs of the ridge lines with the highest matching degree are determined to obtain a first matching result; the first alignment condition is that the starting ridge line point and / or the ending ridge line point of the ridge line is formed by a non-linear interruption or a linear bifurcation; When the ridge line meets the second alignment condition, the ridge line is positioned at the inflection point according to the curvature feature, and the ridge line is matched with the line coincidence and rigid transformation by using the inflection point obtained by positioning, and several pairs of ridge lines with the highest matching degree are determined to obtain a second matching result; the second alignment condition is that the maximum curvature of the ridge line is greater than a preset maximum curvature threshold; When the ridge lines do not meet the first alignment condition and the second alignment condition, rigid transformation matching and transformation angle estimation are performed on the ridge lines according to the curvature feature and the line width feature, and line coincidence matching is performed after the ridge lines are rotated to the estimated transformation angle, and several pairs of ridge lines with the highest matching degree are determined to obtain a third matching result; According to the first matching result, the second matching result or the third matching result, aligning the two ridge line graphs to form the alignment area; The similarity matching and rigid transformation matching are performed on the ridge lines according to the line length features, and a plurality of pairs of ridge lines with the highest matching degree are determined to obtain a first matching result, including: Calculating the line length deviation between the ridge lines according to the line length feature; When the line length deviation is less than a preset line length deviation threshold, similarity matching is performed on the ridge descriptors of the ridge lines to obtain descriptor similarity; When the similarity of the descriptors is greater than a preset similarity threshold, rigid transformation matching is performed on the ridge line, and the rotation matrix and translation parameters of the ridge line are solved by the least square method; When the rotation matrix is ​​smaller than a preset rotation threshold and the translation parameter is smaller than a preset translation threshold, selecting a plurality of pairs of ridge lines whose rotation matrix is ​​closest to the preset rotation threshold and whose translation parameter is closest to the preset translation threshold as the first matching results; And / or locating the inflection point of the ridge line according to the curvature feature, performing line coincidence matching and rigid transformation matching on the ridge line using the inflection point obtained by locating, determining several pairs of ridge lines with the highest matching degree, and obtaining a second matching result, including: Determining the maximum curvature of the ridge line according to the curvature feature; When the maximum curvature is greater than a preset maximum curvature threshold, calculating the maximum curvature deviation between the ridge lines; When the maximum curvature deviation is less than a preset curvature deviation threshold, a chain code derivative integration method is used to locate the inflection point of the ridge line, and the inflection point is located; Calculating the line coincidence matching degree of the ridge line according to the number and relative angle of the inflection points; When the line coincidence matching degree is less than a preset line coincidence threshold, rigid transformation matching is performed on the ridge lines to determine a number of pairs of ridge lines with the highest matching degree, and obtain a second matching result; And / or the step of performing rigid transformation matching and transformation angle estimation on the ridge line according to the curvature feature and the line width feature, rotating the ridge line to the estimated transformation angle and then performing line coincidence matching, determining several pairs of the ridge lines with the highest matching degree, and obtaining a third matching result, including: Calculating the average curvature and average line width of the ridge line according to the curvature feature and the line width feature; When the average curvature is greater than a preset average curvature threshold and the average line width is greater than a preset average line width threshold, rigid transformation matching is performed on the ridge line, a rotation matrix of the ridge line is solved by a least squares method, and a translation parameter of the ridge line is calculated by a distance field matching method; When the translation parameter is less than the preset translation threshold, a plurality of pairs of ridge lines whose translation parameters are closest to the preset translation threshold are selected as the third matching results.

2. The small-area fingerprint image matching method according to claim 1, characterized in that: The ridge point features include ridge point category features and ridge point association features. The ridge point features are extracted from the ridge to obtain the ridge point features, including: Extracting ridge points in the ridge map; Classifying the ridge points according to the neighborhood colors of the ridge points and the positions in the ridge map to obtain the ridge point category features; According to the ridge point category feature, the ridge line is tracked to obtain a tracking result, and according to the tracking result, the ridge line and the ridge point are associated to obtain the ridge point association feature.

3. The small-area fingerprint image matching method according to claim 1, characterized in that: The ridge line features include line length features, curvature features, line width features and line type features. The ridge line features are extracted to obtain the ridge line features, including: Performing Freeman chain code encoding on the ridge to obtain a ridge chain code representation; Performing a mapping accumulation operation on the ridge chain code representation to obtain the line length feature; Calculating the curvature of the ridge line according to the relative positions of the ridge line points on the ridge line to obtain the curvature feature; Taking the sampling points on the ridge line as a reference, performing line width detection on the ridge line in a direction perpendicular to the ridge line to obtain the line width feature; According to the target ridge point feature, the ridge line is classified to obtain the line type feature; the target ridge point feature is the ridge point feature of the ridge point located at the beginning and end positions of the ridge line; When the line feature characterizes that the ridge is a target ridge category, the ridge is sampled at intervals, and a corresponding ridge descriptor is generated according to the interval sampling result.

4. The small-area fingerprint image matching method according to claim 1, characterized in that: According to the overall position deviation of the ridge points in the alignment area, or the local position deviation of the local position of the ridge points in the alignment area with the highest degree of alignment overlap when the overall position deviation of the ridge points in the alignment area exceeds a preset deviation range, a ridge point matching operation is performed on the two ridge line graphs to obtain a ridge point matching result, including: Calculate the overall position deviation of the ridge points in the alignment area; Determining whether the overall position deviation exceeds a preset deviation range; If it does not exceed the limit, the ridge point matching result through which the ridge point matching passes is output; If exceeded, when the local position deviation of the ridge point is less than the preset position deviation, a reference area including the ridge point is divided, and one of the two ridge maps is image-blocked to obtain a plurality of ridge map blocks; Recursively inferring each ridge map block based on the reference area to obtain a target mapping transformation relationship; the target mapping transformation relationship is a rigid transformation relationship that minimizes the mapping position deviation between the ridge point in the reference area or in the ridge map block and the ridge point in the mapped ridge map after the ridge point is mapped to the other of the two ridge maps; When the mapping position deviation is less than a preset mapping deviation threshold, a ridge point matching result through which the ridge point matching passes is output.

5. The small-area fingerprint image matching method according to claim 4, characterized in that: The recursive derivation of each ridge line image block based on the reference area to obtain a target mapping transformation relationship includes: Constructing a mapping-deviation problem; the mapping-deviation problem describes the subordinate relationship between the mapping position deviation and the mapping transformation relationship; Calculate the rigid transformation relationship between the ridge points corresponding to the two ridge lines in the reference area to obtain the target mapping transformation relationship of the reference area; Taking the reference area as the initial block, a block recursive operation is performed; the block recursive operation is to use the target mapping relationship of the adjacent blocks as the initial solution of the current ridge block about the mapping-deviation problem in a preset recursive direction, and solve the mapping transformation relationship that minimizes the mapping position deviation. As the operation of the target mapping transformation relationship, the adjacent blocks are the ridge blocks or the reference area adjacent to the current ridge block in the recursive direction.

6. A small-area fingerprint image matching device, characterized in that: include: The first module is used to obtain at least two fingerprint images to be matched; The second module is used to pre-process the fingerprint image to be matched to obtain a ridge line map containing ridge lines; The third module is used to extract ridge point features from the ridge to obtain ridge point features; The fourth module is used to extract the line features of the ridge line to obtain the line features of the ridge line; A fifth module is used to perform a ridge alignment operation on the two ridge maps according to the ridge point features and the ridge body features to obtain an alignment area; the alignment area is an area where the two ridge maps partially overlap; A sixth module is used to perform a ridge point matching operation on the two ridge line graphs according to the overall position deviation of the ridge line points in the alignment area, or the local position deviation of the local position of the ridge line points in the alignment area with the highest degree of alignment overlap when the overall position deviation of the ridge line points in the alignment area exceeds a preset deviation range, to obtain a ridge point matching result; The seventh module is used to obtain the matching results of the two fingerprint images to be matched according to the ridge point matching results; The ridge line body features include line length features, curvature features, line width features and line type features. The ridge line alignment operation is performed on the two ridge line graphs according to the ridge line point features and the ridge line body features to obtain an alignment area, including: When the ridge line meets the first alignment condition, similarity matching and rigid transformation matching are performed on the ridge line according to the line length feature, and several pairs of the ridge lines with the highest matching degree are determined to obtain a first matching result; the first alignment condition is that the starting ridge line point and / or the ending ridge line point of the ridge line is formed by a non-linear interruption or a linear bifurcation; When the ridge line meets the second alignment condition, the ridge line is positioned at the inflection point according to the curvature feature, and the ridge line is matched with the line coincidence and rigid transformation by using the inflection point obtained by positioning, and several pairs of ridge lines with the highest matching degree are determined to obtain a second matching result; the second alignment condition is that the maximum curvature of the ridge line is greater than a preset maximum curvature threshold; When the ridge lines do not meet the first alignment condition and the second alignment condition, rigid transformation matching and transformation angle estimation are performed on the ridge lines according to the curvature feature and the line width feature, and line coincidence matching is performed after the ridge lines are rotated to the estimated transformation angle, and several pairs of ridge lines with the highest matching degree are determined to obtain a third matching result; According to the first matching result, the second matching result or the third matching result, aligning the two ridge line graphs to form the alignment area; The similarity matching and rigid transformation matching are performed on the ridge lines according to the line length features, and a plurality of pairs of ridge lines with the highest matching degree are determined to obtain a first matching result, including: Calculating the line length deviation between the ridge lines according to the line length feature; When the line length deviation is less than a preset line length deviation threshold, similarity matching is performed on the ridge descriptors of the ridge lines to obtain descriptor similarity; When the similarity of the descriptors is greater than a preset similarity threshold, rigid transformation matching is performed on the ridge line, and the rotation matrix and translation parameters of the ridge line are solved by the least square method; When the rotation matrix is ​​smaller than a preset rotation threshold and the translation parameter is smaller than a preset translation threshold, selecting a plurality of pairs of ridge lines whose rotation matrix is ​​closest to the preset rotation threshold and whose translation parameter is closest to the preset translation threshold as the first matching results; And / or locating the inflection point of the ridge line according to the curvature feature, performing line coincidence matching and rigid transformation matching on the ridge line using the inflection point obtained by locating, determining several pairs of ridge lines with the highest matching degree, and obtaining a second matching result, including: Determining the maximum curvature of the ridge line according to the curvature feature; When the maximum curvature is greater than a preset maximum curvature threshold, calculating the maximum curvature deviation between the ridge lines; When the maximum curvature deviation is less than a preset curvature deviation threshold, a chain code derivative integration method is used to locate the inflection point of the ridge line, and the inflection point is located; Calculating the line coincidence matching degree of the ridge line according to the number and relative angle of the inflection points; When the line coincidence matching degree is less than a preset line coincidence threshold, rigid transformation matching is performed on the ridge lines to determine a number of pairs of ridge lines with the highest matching degree, and obtain a second matching result; And / or the step of performing rigid transformation matching and transformation angle estimation on the ridge line according to the curvature feature and the line width feature, rotating the ridge line to the estimated transformation angle and then performing line coincidence matching, determining several pairs of the ridge lines with the highest matching degree, and obtaining a third matching result, including: Calculating the average curvature and average line width of the ridge line according to the curvature feature and the line width feature; When the average curvature is greater than a preset average curvature threshold and the average line width is greater than a preset average line width threshold, rigid transformation matching is performed on the ridge line, a rotation matrix of the ridge line is solved by a least squares method, and a translation parameter of the ridge line is calculated by a distance field matching method; When the translation parameter is less than the preset translation threshold, a plurality of pairs of ridge lines whose translation parameters are closest to the preset translation threshold are selected as the third matching results.

7. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the small-area fingerprint image matching method according to any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the small-area fingerprint image matching method according to any one of claims 1 to 5 is implemented.