Method for utilizing local features of orb image to reduce fingerprint false acceptance·

Abstract

Disclosed in the present invention is a method for utilizing local features of an ORB image to reduce fingerprint false acceptance. The method comprises: S1, extracting detail points of fingerprint images to be matched; S2, marking the surrounding areas of the detail points of the fingerprint images to be matched, and extracting ORB key points of the marked areas; S3, calculating matched detail point pairs of the fingerprint images to be matched, determining whether the calculated matched detail point pairs are within a preset detail point pair range or not, if not, failing to match the fingerprint images to be matched, and if yes, executing step S4; S4, on the basis of the calculated matched detail point pairs, matching the ORB key points of the fingerprint images to be matched, determining whether the number of the matched ORB key points is greater than or equal to the number of preset matched ORB key points, if not, failing to match the fingerprint images to be matched, and if yes, executing step S5; and S5, successfully matching the fingerprint images to be matched.

Description

A method to reduce false fingerprint recognition by using local features of ORB images Technical Field

[0001] The present invention relates to the technical field of fingerprint matching, and in particular to a method for reducing false fingerprint recognition by utilizing local features of ORB images. Background Art

[0002] Fingerprint recognition terminals currently on the market are using various methods to reduce hardware costs and expand their product applications, the most prominent of which is the reduction in sensor size. The image size captured by fingerprint sensors at the same DPI has been reduced from 256*360 to 160*160. The captured area no longer covers the entire finger, and as the capture area decreases, the number of minutiae unique to fingerprint patterns also decreases. Minutiae points on different fingers are often similar, and matching algorithms based solely on these minutiae can result in numerous false positives.

[0003] Patent publication number CN115497125A discloses a fingerprint recognition method, system, computer device, and computer-readable storage medium, including: obtaining a fingerprint image of a fingerprint to be matched; extracting minutiae from the fingerprint image to obtain minutiae features; extracting and compressing the minutiae depth features of the minutiae features to obtain vector descriptors of the minutiae of the fingerprint image to be matched; using the vector descriptors to pattern-match the minutiae features of the fingerprint image to be matched with the minutiae features of the template fingerprint image; if the match is successful, the match is complete; if the match fails, the fingerprint image to be matched is verified by ORB feature matching with the template fingerprint image. Although the above patent can achieve accurate matching on smaller fingerprint image sizes, its verification method adjusts the minutiae feature matching score, and false recognition may still occur in fingerprint minutiae matching.

[0004] In response to the above technical problems, the present invention provides a method for reducing false fingerprint recognition by utilizing local features of ORB images.

[0005] Summary of the Invention

[0006] The purpose of the present invention is to address the defects of the prior art and provide a method for reducing false fingerprint recognition by utilizing local features of ORB images.

[0007] In order to achieve the above objectives, the present invention adopts the following technical solutions:

[0008] A method for reducing false fingerprint recognition by using local features of ORB images, comprising:

[0009] S1. Extract the minutiae of the fingerprint image to be matched;

[0010] S2. Mark the area around the minutiae of the fingerprint image to be matched and extract the ORB key points of the marked area;

[0011] S3 calculates the minutiae matching point pairs of the fingerprint image to be matched, and determines whether the minutiae matching point pairs calculated are within the preset minutiae matching point range. If not, the matching fingerprint image fails; if so, proceed to step S4;

[0012] S4 is matched according to the calculated minutiae matching points to be matched against the ORB key points of the fingerprint image, and it is determined that the number of ORB key point matches is greater than or equal to the preset number of ORB key point matches, if otherwise the fingerprint image to be matched fails; if so, proceed to step S5;

[0013] S5. The fingerprint image to be matched is matched successfully.

[0014] Furthermore, after extracting the minutiae points of the fingerprint image to be matched in step S1, the method further includes: calculating the distances between all minutiae points and the center point of the fingerprint image to be matched, and sorting the minutiae points corresponding to the calculated distances.

[0015] Furthermore, the area marked in step S2 is as follows: the original image of the fingerprint image to be matched is divided into blocks, and blocks around the minutiae that are not near the background area are selected, and the selected blocks are marked to obtain a marked area.

[0016] Furthermore, the step S2 of extracting the ORB key points of the marked area includes removing the ORB key points that are too close in the marked area and selecting the ORB key points that are close to the minutiae of the fingerprint image to be matched.

[0017] Furthermore, in step S3, the minutiae matching point pairs of the to-be-matched fingerprint image are calculated, specifically by obtaining minutiae features of the to-be-matched fingerprint image, and matching the minutiae features of the to-be-matched fingerprint image to obtain a minutiae matching result.

[0018] Furthermore, in step S4, the ORB key points of the fingerprint image to be matched are matched according to the calculated minutiae matching points, specifically:

[0019] A1. Obtain ORB key point features of the fingerprint image to be matched;

[0020] A2. Convert the ORB key point features of the fingerprint image to be matched to a unified matching coordinate for registration;

[0021] A3. Match the ORB key point features of the registered fingerprint image to be matched and obtain the matching result.

[0022] Furthermore, the detail point features include the coordinate position of the detail point and the direction of the detail point.

[0023] Furthermore, the ORB key point features include the coordinate position of the ORB feature point, the direction of the ORB feature point, and the ORB feature point descriptor.

[0024] Furthermore, before step A2, the method further includes: calculating configuration parameters of ORB key point features.

[0025] Compared with the prior art, the present invention does not extract ORB features from all areas, but only extracts ORB features from areas surrounding minutiae. At the same time, an ORB feature screening method is provided, which can flexibly control the number of ORB feature points as needed, thereby reducing the operation time and the size of the fingerprint feature file. The present invention increases ORB features, thereby reducing the possibility of false recognition and mismatching of minutiae points due to similar topological structures. The ORB feature matching algorithm of the present invention makes full use of minutiae matching pairs, simplifies the entire ORB feature matching process, and reduces additional operation time. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] FIG1 is a flow chart of a method for reducing false fingerprint recognition by using local features of an ORB image, provided in Example 1;

[0027] FIG2 is a schematic diagram of the number of ORB feature points extracted using the conventional method provided in Example 1;

[0028] FIG3 is a schematic diagram of fingerprint minutiae provided in Example 1;

[0029] FIG4 is a schematic diagram of a block screened for ORB feature extraction provided in Example 1;

[0030] FIG5 is a schematic diagram of ORB features and fingerprint minutiae of the filtered blocks provided in Example 1;

[0031] FIG6 is a schematic diagram of the same finger matching situation provided in Example 1;

[0032] FIG7 is a schematic diagram of the matching of different fingers provided in the second embodiment. DETAILED DESCRIPTION

[0033] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other unless they conflict.

[0034] The purpose of the present invention is to address the defects of the prior art and provide a method for reducing false fingerprint recognition by utilizing local features of ORB images.

[0035] Example 1

[0036] This embodiment provides a method for reducing false fingerprint recognition by using local features of ORB images, as shown in FIG1 , including:

[0037] S1. Extract the minutiae of the fingerprint image to be matched;

[0038] S2. Mark the area around the minutiae of the fingerprint image to be matched and extract the ORB key points of the marked area;

[0039] S3 calculates the minutiae matching point pairs of the fingerprint image to be matched, and determines whether the minutiae matching point pairs calculated are within the preset minutiae matching point range. If not, the matching fingerprint image fails; if so, proceed to step S4;

[0040] S4 is matched according to the calculated minutiae matching points to be matched against the ORB key points of the fingerprint image, and it is determined that the number of ORB key point matches is greater than or equal to the preset number of ORB key point matches, if otherwise the fingerprint image to be matched fails; if so, proceed to step S5;

[0041] S5. The fingerprint image to be matched is matched successfully.

[0042] The present invention studies various local image feature methods such as SIFT, SURF, and ORB. Among them, ORB has a small amount of calculation and its descriptor is binary coded, which has a relatively small storage capacity and is rotationally invariant. It is very suitable for use in low-resource fingerprint recognition embedded systems. However, a fingerprint image of medium area (160*160) has rich lines and textures, and the number of extracted ORB local feature points is large, which takes up a lot of storage space. As shown in Figure 2, there are 550 ORB feature points in total. It is extremely unreasonable to store all of them in the fingerprint template for comparison. In this regard, this embodiment proposes a method for reducing fingerprint false recognition by utilizing local features of ORB images to solve the above problem.

[0043] In step S1, minutiae points of the fingerprint image to be matched are extracted.

[0044] Obtain the fingerprint images to be matched and extract the minutiae (endpoints and cross points) of the fingerprint images. The fingerprint images to be matched are the first fingerprint image and the second fingerprint image.

[0045] It should be noted that the method for extracting minutiae from the fingerprint image may adopt an existing solution, which will not be described in detail in this embodiment.

[0046] As shown in Figure 3, the black dots are erroneous detail points caused by image breaks, and the white dots are image detail points. In this embodiment, if the number of extracted detail points is greater than 30 (the number of detail points can be adjusted according to the maximum space that can be stored in the actual template), ORB feature extraction is not performed on all detail points, and the distance between the extracted detail points and the image center point is calculated and sorted.

[0047] In subsequent operations, this embodiment does not perform ORB keypoint extraction on the central minutiae area. Instead, it extracts ORB keypoints only for minutiae away from the center of the image. This is because minutiae away from the center are close to the edge of the image. During registration, the overlapping area of ​​the fingerprint image feature points to be matched will include the edge area. The smaller the overlapping area, the less central area is included, and the chance of ORB matching in the central area decreases. By controlling the number of minutiae, the extraction area of ​​ORB feature points can be controlled.

[0048] In step S2, the area around the minutiae of the fingerprint image to be matched is marked, and the ORB key points of the marked area are extracted.

[0049] The original fingerprint image to be matched is divided into 8x8 pixel blocks. The blocks are then checked for fingerprint texture. If no texture is present, they are considered background. Blocks close to the background are prone to error and are therefore not selected. The final ORB region is the area surrounding the selected minutiae (5x5 pixels) and not near the background. As shown in Figure 4, white dots represent marked blocks, and black dots represent image minutiae. The selected blocks are marked, and ORB keypoints are extracted from the marked areas. Because fingerprint images are not scaled, corner detection can be performed using the ORB keypoint detection method (FAST) at the original image scale, reducing computational complexity.

[0050] In this embodiment, ORB keypoint calculations are performed on the filtered area. If there are still too many keypoints, resulting in a large feature file, ORB keypoints that are too close to each other can be removed, ORB keypoints that are close to the fingerprint minutiae can be selected, and ORB keypoints with the highest quality can be selected. As shown in Figure 5, white dots are ORB keypoints, black dots are fingerprint minutiae, and black lines are ORB directions. The number of feature points is reduced from 550 to 75. Their directions and feature descriptors are calculated and stored in the feature template along with the fingerprint minutiae information (endpoints and cross points).

[0051] Existing ORB keypoint matching algorithms generally first perform a 1:1 search on the descriptors of features A and B to find keypoint matching pairs with high similarity. They then use the RANSAC iterative method to estimate the matching relationship between features A and B. The RANSAC iterative method: Based on the descriptor matching, a series of possible ORB matching pairs for features A and B are found. Some of these possible ORB matching pairs are randomly selected as initial inliers. A registration relationship model is then fitted using these inliers. All other possible ORB matching pairs are then tested. If a possible ORB matching pair fits this registration relationship model, it is considered an inlier and expanded. The number of possible ORB matching pairs used as RANSAC inliers is calculated, and the error rate of the inlier point registration relationship model is estimated. This entire process is repeated several times (iterations N > 10). Each iteration generates a new registration relationship model. If it is inferior to the previous model due to too few inliers or a high error rate, it is discarded. If it is better than the existing model, it is selected. The best registration relationship model is selected to remove incorrect ORB matching pairs. The ORB keypoint matching algorithm is relatively complex and requires a large amount of computation. Therefore, this embodiment redesigns a new ORB minutiae matching method, as shown in steps S3-S4.

[0052] In step S3, the minutiae matching point pairs of the to-be-matched fingerprint image are calculated, and it is determined whether the calculated minutiae matching point pairs are within the preset minutiae point pair range. If not, the matching of the to-be-matched fingerprint image fails; if so, step S4 is executed.

[0053] The minutiae points of the first fingerprint image and the second fingerprint image are compared to obtain matching point pairs of the minutiae points of the first fingerprint image and the second fingerprint image features. Based on these point pairs, the registration relationship between the first fingerprint image and the second fingerprint image features is calculated to obtain the relevant matching situation of ORB. This matching method is fast and has a small amount of calculation. When the number of matching minutiae point pairs is greater than or equal to 3 pairs and less than 15 pairs, the fingers are considered to be the same finger and step S4 is executed.

[0054] As shown in Figure 6, fingerprint images of two identical fingers are collected. The white dots are ORB key feature points, the black dots are fingerprint detail points, and the black lines are ORB directions. This embodiment is illustrated using feature point A of the first fingerprint image and feature point B of the second fingerprint image.

[0055] Feature extraction is performed based on the image to obtain minutiae feature information A of the first fingerprint image and minutiae feature information B of the second fingerprint image. The minutiae feature information includes the coordinate position x, y and direction ang of the minutiae.

[0056] Minutiae points a1, a2, and a3 are points that match minutiae point A obtained by a minutiae point matching algorithm; minutiae points b1, b2, and b3 are points that match minutiae point B obtained by a minutiae point matching algorithm.

[0057] From this we can see that a1 (x = 76, y = 141, ang = 240) and b1 (x = 27, y = 120, ang = 231), a2 (x = 94, y = 77, ang = 270) and b2 (x = 39, y = 54, ang = 265), a3 (x = 115, y = 56, ang = 310) and b3 (x = 58, y = 30, ang = 305) match.

[0058] It should be noted that the minutiae matching algorithm may adopt an existing solution, which will not be described in detail in this embodiment.

[0059] According to the minutiae matching results, calculate the translation and rotation factors. Taking minutiae point A as the reference, rotate and translate feature B to the coordinate system of feature A for comparison, and match the minutiae point according to the minutiae point matching results. a1 and b1 are the matched reference pairs, and the registration parameters are TransParaB (baseX, baseY, deltaX, deltaY, deltaTheta), where baseX is the horizontal coordinate of the reference point, baseY is the vertical coordinate of the reference point, deltaX is the horizontal coordinate offset, deltaY is the vertical coordinate offset, and deltaTheta is the angle offset. Take the minutiae point reference pair b1 in feature B as the reference point and calculate the configuration parameters TransParaB, specifically:

[0060] According to the above calculation method, TransParaB (baseX=27, baseY=120, deltaX=49, deltaY=21, deltaTheta=351) can be obtained.

[0061] In step S4, the ORB key points of the fingerprint image to be matched are matched according to the calculated minutiae matching points, and the number of ORB key point matches is determined to be greater than or equal to the preset ORB key point matching number. If not, the matching of the fingerprint image to be matched fails; if so, step S5 is executed; specifically including:

[0062] A1. Obtain the ORB key point features of the fingerprint image to be matched.

[0063] Feature extraction is performed based on the image to obtain ORB feature A of the first fingerprint image and ORB feature B of the second fingerprint image, wherein the ORB feature includes the coordinate position x, y of the ORB feature point, the direction ang of the ORB feature point, and the ORB feature descriptor binary.

[0064] During the ORB keypoint comparison phase, the ORB keypoints in the minutiae area that match the position are matched for direction and descriptor matching. If there are five or more ORB keypoints that match, the minutiae and keypoints are considered matched successfully, and the probability of false recognition is extremely low. If the two finger images are of the same finger, the match is successful; otherwise, the two fingers are considered different and the match fails.

[0065] A2. Convert the ORB key point features of the fingerprint image to be matched to a unified matching coordinate for registration.

[0066] After obtaining the configuration parameter TransParaB in the detail point calculation in step S3, the ORB feature (d1, d2, d3...) in the B feature is rotated and translated to the A feature coordinate system according to the calculated registration parameters. The new horizontal coordinate, new vertical coordinate, and new angle of the ORB feature are Newdx, Newdy, and Newdang. The binary descriptor is rotation-translation invariant and does not need to be changed. The specific implementation method is as follows:

[0067] The ORB feature point d1 (x=102, y=116, ang=351, binary=1001111101000001 10110111) in the B feature is translated and rotated to the A feature coordinate system according to the calculated TransParaB (baseX=27, baseY=120, deltaX=49, deltaY=21, deltaTheta=351) configuration parameters to obtain d11. The specific calculation details are as follows: SinTheta=sin(351)=-0.156 CosTheta=cos(351)=0.99 d11(x)=(int)((27+49)+(116-120)*(-0.156)+(102-27)*0.99+0.5)=151; d11(y)=(int)((120+21)+(116-120)*0.99-(102-27)*(-0.156)+0.5)=149; d11(ang)=351+351-360=342 d11(binary)=d1(binary)=10011111 01000001 10110111;

[0068] Calculation yields d11(x=151,y=149,ang=342,binary=10011111 01000001 10110111).

[0069] A3. Match the ORB key point features of the registered fingerprint image to be matched and obtain the matching result.

[0070] Based on the coordinates of d11, we can find that point c1 in feature A matches it, and the error range is within the allowable threshold. Δx is between -5 and +5, Δy is between -5 and +5, and Δang is between -15 and +15. Δbinary has a maximum of five dimensions that are inconsistent in the 24-dimensional binary data. The error values ​​between c1 and d11 are as follows. They are within the error range and are therefore a match. Δx=d11(x)-c1(x)=151-153=-2; Δy=d11(y)-c1(y)=149-146=3; Δang=d11(ang)-c1(ang)=342-350=-8; Δbinary=d11(10011111 01000001 10110111)-c1(10111111 11001011 10110111)=3

[0071] Similarly, all other ORB features in B are rotated and translated to the A feature coordinate system in the same way as above. It can be calculated that a total of ten pairs of points are matched, and the error is within the allowable range. 0100001 10101011) c2(x=134,y=97,ang=253,binary=00110111 00100001 10101011) d1(x=102,y=116,ang=351,binary=10011111 01000001 10110111) d2(x=134,y=97,ang=253,binary=00110111 00100001 10101011) d2(x=80,y=68,ang=233,binary=01110111 00100001 00110011) c3(x=82,y=122,ang=298,binary=00111110 00100000 10100101) d3(x=30,y=99,ang=276,binary=00111110 10000100 11100100) c4(x=93,y=58,ang=305,binary=00011111 11000001 01110111) d4(x=36,y=34,ang=306,binary=00011101 11000000 01110101) c5(x=80,y=145,ang=36,binary=00100000 10110100 11000100) d5(x=31,y=123,ang=35,binary=00101110 10111110 11100100) c6(x=93,y=150,ang=186,binary=11000100 00000000 101000) d6(x=44,y=125,ang=186,binary=00000100 00000000 101000) c7(x=71,y=158,ang=46,binary=01001000 10010000 01000010) d7(x=25,y=138,ang=29,binary=00001000 10011010 11111010) c8(x=82,y=83,ang=132,binary=01000001 01011011 11000001) d8(x=27,y=57,ang=133, binary=01000001 10011010 11100000) c9(x=98, y=85, ang=211, binary=01000001 01011011 01100000) d9(x=43, y=61, ang=196, binary=01000101 01011011 01101000) c10(x=85,y=114,ang=292,binary=00011111 01000001 11110111) d10(x=32,y=91,ang=271,binary=01011111 00000000 11110100),

[0072] According to the above content, all the A feature and B feature detail points are matched successfully, and the ORB feature is matched successfully, which means it is the same finger.

[0073] This embodiment uses the image detail feature point as the center point to retain the required local ORB feature points, which occupies very little feature template storage space. At the same time, the comparison method is performed through the ORB local feature to reduce the false recognition rate of the medium-area fingerprint detail point comparison algorithm.

[0074] Example 2

[0075] The method for reducing false fingerprint recognition by using local features of ORB images provided in this embodiment differs from that in the first embodiment in that:

[0076] In this embodiment, feature A of the first fingerprint image and feature C of the second fingerprint image are compared.

[0077] As shown in Figure 7, the white dots are ORB key feature points, the black dots are fingerprint minutiae, and the black lines are ORB direction indicators. Feature A and Feature C are from different fingers. However, existing minutiae matching algorithms can successfully match them, thus resulting in a false positive.

[0078] According to this solution, the detail points of feature A are e1, e2, and e3, and the detail points of feature C are f1, f2, and f3. Then:

[0079] e1(x=136,y=89,ang=268) and f1(x=93,y=79,ang=270), e2(x=150,y=143,ang=175) and f2(x=97,y=128,ang=177), e3(x=80,y=73,ang=267) and f3(x=41,y=58,ang=268) are matching pairs in detail point matching.

[0080] However, the ORB pairing was not successful, specifically:

[0081] According to the method of step S4, e1 and f1 are used as the reference pair to obtain the registration parameters TransParaC (baseX=93, baseY=79, deltaX=43, deltaY=10, deltaTheta=2). According to the registration parameters TransParaC, the ORB features of feature C are registered to the coordinate system of feature A. No ORB features are registered with the ORB features in feature A. For example:

[0082] g1(x=129,y=95,ang=344,binary=01001001 01000001 10101001),g2(x=80,y=145,ang=36,binary=00100000 10110100 11000000),g3(x=71,y=158, ang=46,binary=01001000 10010000 1000010) are the ORB feature points of feature A,h1(x=86,y=85,ang=354,binary=00001000 00010000 10000100),h2(x=32,y=133,ang=45,binary=11110101 01110001 00011001), h3 (x=35, y=161, ang=295, binary=101111110 11100101 11111100) are the ORB feature points of feature C. The translation and rotation results of f1, f2, and f3 to the A feature coordinate system are calculated according to the registration parameter TransParaC, and the obtained results are g11(x=129, y=95, ang=356, binary=00001000 00010000 10000100), g22(x=76, y=145, ang=47, binary=11110101 01110001 00011001), and g33(x=80, y=172, ang=297, binary=101111110 11100101 11111100).

[0083] The coordinate positions and angles of g1 (x=129, y=95, ang=344, binary=01001001 01000001 10101001) and g11 (x=129, y=95, ang=356, binary=00001000 00010000 10000100) are within the error range, but the binary descriptors have 9 different dimensions, which exceeds the allowable error range.

[0084] The coordinate positions and angles of g2 (x = 80, y = 145, ang = 36, binary = 00100000 10110100 11000000) and g22 (x = 76, y = 145, ang = 47, binary = 11110101 01110001 00011001) are within the error range, but the binary descriptors have 14 different dimensions, which exceeds the allowable error range.

[0085] For g3 (x=71, y=158, ang=46, binary=01001000 10010000 1000010) and g33 (x=80, y=172, ang=297, binary=101111110 11100101 11111100), the coordinate position and angle binary descriptors are both far beyond the error range.

[0086] Therefore, the ORB key point feature matching of feature A and feature C is unsuccessful. Feature A and feature C are not the same finger, and the comparison fails.

[0087] Compared with the algorithm that only relies on minutiae points, the ORB feature designed in this embodiment can effectively reduce the false recognition rate of the fingerprint recognition algorithm.

[0088] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for reducing false fingerprint recognition by using local features of ORB images, characterized in that: include: S1. Extract the minutiae of the fingerprint image to be matched; S2. Mark the area around the minutiae of the fingerprint image to be matched, and extract the ORB key points of the marked area; S3. Calculate the minutiae matching point pairs of the fingerprint image to be matched, and determine whether the calculated minutiae matching point pairs are within the preset minutiae matching point range. If not, the matching of the fingerprint image to be matched fails; If yes, execute step S4; S4. Match the ORB key points of the fingerprint image to be matched according to the calculated minutiae matching points, and determine that the number of ORB key point matches is greater than or equal to the preset number of ORB key point matches, if otherwise the fingerprint image to be matched fails to match; if so, execute step S5; S5. The fingerprint image to be matched is matched successfully.

2. The method of reducing false fingerprint recognition by using local features of ORB images according to claim 1, characterized in that: After extracting the minutiae points of the fingerprint image to be matched in step S1, the step further includes: calculating the distances between all minutiae points and the center point of the fingerprint image to be matched, and sorting the minutiae points corresponding to the calculated distances.

3. The method of reducing false fingerprint recognition by using local features of ORB images according to claim 1, characterized in that: The area marked in step S2 is: the original image of the fingerprint image to be matched is divided into blocks, and blocks around the detail points and not near the background area are selected, and the selected blocks are marked to obtain the marked area.

4. The method of reducing false fingerprint recognition by using local features of ORB images according to claim 3, characterized in that: The step S2 of extracting the ORB key points of the marked area includes removing the ORB key points that are too close to the marked area and selecting the ORB key points that are close to the minutiae of the fingerprint image to be matched.

5. The method of reducing false fingerprint recognition by using local features of ORB images according to claim 4, characterized in that: The step S3 calculates the detail point matching point pairs of the fingerprint image to be matched, specifically: obtaining the detail point features of the fingerprint image to be matched, and matching the detail point features of the fingerprint image to be matched to obtain the detail point feature matching results.

6. The method of reducing false fingerprint recognition by using local features of ORB images according to claim 5, characterized in that: In step S4, the calculated minutiae matching points are matched with the ORB key points of the fingerprint image to be matched, specifically: A1. Obtain the ORB key point features of the fingerprint image to be matched; A2. Convert the ORB key point features of the fingerprint image to be matched to a unified matching coordinate for registration; A3. Match the ORB key point features of the registered fingerprint image to be matched and obtain the matching result.

7. The method of reducing false fingerprint recognition by using local features of ORB images according to claim 5, characterized in that: The detail point features include the coordinate position of the detail point and the direction of the detail point.

8. The method of reducing false fingerprint recognition by using local features of ORB images according to claim 6, characterized in that: The ORB key point features include the coordinate position of the ORB feature point, the direction of the ORB feature point, and the ORB feature point descriptor.

9. The method of reducing false fingerprint recognition by using local features of ORB images according to claim 6, characterized in that: Before step A2, the method further includes: calculating configuration parameters of ORB key point features.

Images