Fast fingerprint image binaryzation and feature extraction method

Abstract

The invention discloses a fast fingerprint image binaryzation and feature extraction method based on a pre-integrogram and a gradient direction. The problems that an existing fingerprint image binaryzation and feature extraction method is high in calculation complexity and poor in real-time performance are mainly solved. The method mainly comprises the steps of performing pre-integration on an input fingerprint image to obtain a pre-integral graph; calculating the ridge line direction of the position of each pixel point through a gradient method; according to the ridge line direction, the line average value in the ridge line direction is compared with the surface average value obtained through rapid calculation of the pre-integrogram, if the line average value is smaller than the surface average value, it is judged that the fingerprint line is a ridge line, otherwise, it is judged that the fingerprint line is a fingerprint valley line, and rapid binaryzation is achieved; and the binarized image is further refined. According to the method, the pre-integrogram can be utilized to quickly binarize the fingerprint image, so that the fingerprint feature extraction time is greatly shortened, and the method can be used for identity recognition in embedded equipment.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing, and relates to fast binarization and feature extraction of fingerprint images, and in particular to fast binarization of fingerprint images, which can be used for identity authentication in embedded devices. Background Art

[0002] As a biometric feature, fingerprints have been widely used in the fields of identity recognition, access control, attendance, etc. People have conducted a lot of research on this for a long time and have made great progress. However, there are still some problems in the field of fingerprint recognition that need to be studied and solved. Especially in the real-time fingerprint recognition system of embedded platforms, the speed of fingerprint feature extraction is seriously restricted by the computing power of embedded devices, which has become a key issue. Therefore, how to improve the computing speed of fingerprint recognition algorithms has become an important research topic. In the process of fingerprint feature extraction, the binarization process of the image consumes the most computing time, accounting for about 45% of the entire feature extraction process. Therefore, how to reduce the binarization time will greatly reduce the time of the entire feature extraction. The traditional binarization method requires repeated accumulation and averaging of the field of each pixel point, which requires a large amount of calculation, thus greatly increasing the feature extraction time. Summary of the invention

[0003] Aiming at the shortcomings of existing fingerprint image feature extraction methods, the present invention proposes a fast binarization and feature extraction method based on image pre-integration, which avoids a large number of repeated summation calculations, thereby greatly improving the extraction speed of fingerprint features.

[0004] The technical solution adopted by the present invention is as follows:

[0005] A fast fingerprint image binarization and feature extraction method comprises the following steps:

[0006] The input fingerprint image I(x, y) is pre-integrated to obtain a pre-integrated graph G(x, y), where each element value G(x, y) in the pre-integrated graph is the sum of all pixel values ​​in the upper left corner of the pixel (x, y);

[0007] The local ridge direction of each pixel of the fingerprint image is estimated based on the gradient algorithm;

[0008] The input image I(x, y) is quickly binarized according to the ridge direction and the pre-integrated image to obtain a binary image B(x, y), where the value of B(x, y) is 0 or 1;

[0009] Thinning the binary image B(x, y);

[0010] Perform fingerprint feature point detection on the refined image;

[0011] After deleting the pseudo feature points, the fingerprint feature file is obtained.

[0012] In a preferred embodiment, the fingerprint image binarization and feature extraction method includes a pre-integration method, and the pre-integration method is as follows: each element value G (x, y) in the pre-integration graph is calculated according to the following formula:

[0013] G(x, y) = ∑ x′≤x，y′≤y I(x′, y′)

[0014] Where I(x′, y′) is the grayscale value of the pixel of the input fingerprint image (x′, y′).

[0015] In a preferred embodiment, the fingerprint image binarization and feature extraction method includes a gradient-based ridge direction estimation method, and the gradient-based ridge direction estimation method includes the following steps:

[0016] Calculate the gradient of the fingerprint image I(x, y) in and are the gradients in the x and y directions of pixel I(x, y), respectively, where:

[0017]

[0018] Calculate the local ridge direction based on the phase angle of the gradient;

[0019] The direction perpendicular to the gradient is the direction of the ridge:

[0020]

[0021] In a preferred embodiment, the fingerprint image binarization and feature extraction method includes a fingerprint image fast binarization method, and the fingerprint image fast binarization method includes the following steps:

[0022] The steps of fingerprint image binarization are as follows:

[0023] For each pixel I(x, y) in the fingerprint image, take a preset number of 2l+1 pixels along the ridge line with the current pixel as the center and calculate the average value m l (x, y);

[0024] The average value m of the surface in the (2m+1)×(2n+1) neighborhood centered on the current pixel s (x, y), where the average value of the surface is m s (x, y) is obtained from the pre-integrated graph through several additions and subtractions and one division:

[0025]

[0026] Where G(x, y) is the value at coordinate (x, y) in the pre-integration graph;

[0027] Line mean m l The average gray value of (x, y) is less than the average gray value of the surface m s (x, y), then set B(x, y) to 1, which means it is a ridge line, and set B(x, y) to 0, which means it is a valley line.

[0028] In a preferred embodiment, the fingerprint image binarization and feature extraction method includes a binary image thinning method, and the binary image thinning method includes the following steps:

[0029] For each pixel B(x, y) in the binary image, if B(x, y) = 1 and the following three conditions are met, then B(x, y) is set to 0;

[0030] Condition (1) C(p) = 1, indicating that the number of times the pixel p's 8-neighborhood changes from 0 to 1 in the clockwise direction is 1;

[0031] Condition (2) 2≤min[N1(p), N2(p)]≤3, where

[0032] N1(p)=(p1∨p2)+(p3∨p4)+(p5∨p6)+(p7∨p8),

[0033] N2(p)=(p2∨p3)+(p4∨p5)+(p6∨p7)+(p8∨p1)

[0034] Condition (3)

[0035] Where p is the current pixel, p1, ..., p8 are the 8 fields of p in clockwise order starting from the upper left corner;

[0036] The binary image B(x, y) is repeatedly processed according to the method described in step (4) until no pixels are set to 0.

[0037] In a preferred embodiment, the fingerprint image binarization and feature extraction method includes performing a fingerprint feature extraction method on a binarized image, and the fingerprint feature extraction method on a binarized image includes the following steps:

[0038] For each pixel p in the refined image, if p = 1 and: H(p) = 2, then p is an endpoint; H(p) = 6, then p is a bifurcation point, where:

[0039] Where p9 = p1

[0040] p is the current pixel, p1, ..., p8 are the 8 fields of p in clockwise order starting from the upper left corner.

[0041] In a preferred embodiment, the fingerprint image binarization and feature extraction method includes a pseudo feature point deletion method, and the pseudo feature point deletion method includes the following steps:

[0042] For each candidate feature point, if it is close to the edge of the fingerprint area, the feature point is a pseudo feature point and is deleted;

[0043] For any two endpoints of all candidate feature points, if their directions are opposite and their distance is less than the preset value, these two feature points are pseudo feature points and will be deleted.

[0044] Among all the candidate feature points, if the distance from one endpoint to the other two endpoints is less than a preset value, the endpoint is considered to be a pseudo feature point and is deleted.

[0045] In a preferred embodiment, the fingerprint image binarization and feature extraction method includes a pre-integration image optimization method, and the pre-integration image optimization method includes the following steps:

[0046] Pre-integration image size adjustment: adjust the size of the pre-integration image according to the resolution and size of the input fingerprint image;

[0047] Block processing: Divide the pre-integrated image into multiple sub-blocks, and calculate each sub-block independently;

[0048] Boundary processing: In the boundary area of ​​the pre-integration image, mirror filling or constant filling is used.

[0049] In a preferred embodiment, the fingerprint image binarization and feature extraction method includes a gradient algorithm optimization method, and the gradient algorithm optimization method includes the following steps:

[0050] Multi-scale gradient calculation: Calculate the gradient of the fingerprint image at different scales to adapt to fingerprint images with different texture densities;

[0051] Gradient direction smoothing: Smooth the gradient direction using Gaussian filtering or mean filtering;

[0052] Directional consistency correction: In a local area, the direction of the current pixel is corrected according to the gradient direction of adjacent pixels.

[0053] The technical effects achieved by the present invention are:

[0054] The present invention can obtain the pixel average value within a preset range of any pixel of the fingerprint image through several additions and subtractions and one division according to the pre-integration graph, without traversing all pixels for cumulative summation, which greatly reduces the time of image binarization and thus realizes fast feature extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1This is the process of the fast fingerprint feature extraction method of the present invention;

[0056] Figure 2 The value of the pre-integration map (x, y) element G(x, y) is the accumulation of the shadow part of the input image;

[0057] Figure 3 The original fingerprint image (a) and its ridge direction map (b);

[0058] Figure 4 The sum of the pixels in the gray rectangle can be calculated by several additions and subtractions;

[0059] Figure 5 is the original fingerprint image and its binarization image;

[0060] Figure 6 is the 8-neighborhood of pixel p;

[0061] Figure 7 is the result after refinement;

[0062] Figure 8 is the direction of the feature point (the left picture is the endpoint, the right picture is the bifurcation point);

[0063] Fig. 9 It is the feature point after deleting the pseudo feature point. DETAILED DESCRIPTION

[0064] Aiming at the shortcomings of the existing fingerprint image feature extraction method, the present invention proposes a fast binarization and feature extraction method based on image pre-integration.

[0065] Example 1

[0066] The process of this method is as follows Figure 1 As shown:

[0067] 1. Pre-integrate the input fingerprint image I(x, y) to obtain the pre-integration graph G(x, y). Each element value G(x, y) in the pre-integration graph is as follows: Figure 2 The figure shows the sum of the upper left pixel values ​​of the pixel point (x, y), calculated according to the following formula:

[0068] G(x, y) = ∑ x′≤x，y′≤y I(x′, y′)

[0069] Where I(x′, y′) is the grayscale value of the pixel of the input fingerprint image (x′, y′).

[0070] 2. Estimate the local ridge direction of each pixel of the fingerprint image based on the gradient algorithm, including the following steps:

[0071] 2a. Calculate the gradient of the fingerprint image I(x, y) in and point

[0072] are the gradients in the x and y directions of pixel I(x, y), where:

[0073]

[0074] 2b. Calculate the local ridge direction based on the phase angle of the gradient: the direction perpendicular to the gradient direction is the direction of the ridge:

[0075]

[0076] like Figure 3 As shown, the left side is the input fingerprint image, and the right side is the direction estimation result.

[0077] 3. Binarize the input image I(x, y) according to the ridge direction to obtain a binary image B(x, y), where the value of B(x, y) is 0. The binarization steps are as follows:

[0078] 3a. For each pixel I(x, y) in the fingerprint image, take 9 pixels along the ridge line with the current pixel as the center and calculate the average value m l (x, y);

[0079] 3b. If Figure 4 As shown in the figure, the sum of the pixel values ​​in any rectangle in the gray area of ​​the input image I(x, y) is obtained by performing a few additions and subtractions based on the pre-integration matrix diagram:

[0080] A=G(x2,y2)-G(x1,y2)-G(x2,y1)+G(x1,y1)

[0081] The average value m of the face in the 9×9 neighborhood centered on the current pixel s (x, y), where the average value of the surface is m s (x, y) is obtained from the pre-integrated graph through several additions and subtractions and one division:

[0082]

[0083] Where G(x, y) is the value at coordinate (x, y) in the pre-integration graph.

[0084] 3c. Line mean m l The average gray value of (x, y) is less than the average gray value of the surface m s (x, y), then set B(x, y) to 1, that is, it is a ridge line, otherwise set B(x, y) to 0, that is, it is a valley line, such as Figure 5 As shown, the left side is the input fingerprint image, and the right side is the binary image after binarization through the above steps.

[0085] 4. Thin the binary image. The basic steps are as follows: for each pixel B(x, y) in the binary image, if B(x, y) = 1 and the following three conditions are met, set B(x, y) to 0.

[0086] 4a. C(p) = 1, indicating that the number of times the pixel p's 8-neighborhood changes from 0 to 1 in the clockwise direction is 1;

[0087] 4b.2≤min[N1(p),N2(p)]≤3, where

[0088] N1(p)=(p1∨p2)+(p3∨p4)+(p5∨p6)+(p7∨p8),

[0089] N2(p)=(p2∨p3)+(p4∨p5)+(p6∨p7)+(p8∨p1)

[0090] (4c)

[0091] Among them, Figure 6 As shown, p is the current pixel, and p1, ..., p8 are 8 fields of p in clockwise order starting from the upper left corner.

[0092] Repeat the process in step 4 for the binary image B(x, y) until no pixels are set to 0. Figure 7 The figure shows the result of thinning the binary image using the above method.

[0093] 5. Fingerprint feature point extraction, such as Figure 8 The fingerprint feature points shown include endpoints and bifurcation points. For each pixel point p in the refined image, p = 1 and: H(p) = 2, then p is an endpoint; H(p) = 6, then p is a bifurcation point, where:

[0094] Where p9 = p1

[0095] As in step 4, Figure 6 As shown, p is the current pixel, p1, ..., p8 are the 8 fields of p in clockwise order starting from the upper left corner.

[0096] 6. Delete pseudo feature points. Affected by noise and fingerprint cracks, many pseudo feature points will appear in the fingerprint image, which is particularly obvious at the edge of the fingerprint area. The feature points detected initially are screened, and the pseudo feature points are removed and the true feature points are retained. Pseudo feature points are deleted according to the following rules:

[0097] 6a. For each feature point extracted by the method in step 5, if it is close to the edge of the fingerprint area, that is, if any feature point appears in a non-fingerprint area within a radius of 6 pixels, the feature point is a pseudo feature point and is deleted;

[0098] 6b. For the feature points remaining after step 6a, if any two endpoints are in opposite directions and the distance is less than 5 pixels, then these two feature points are pseudo feature points and shall be deleted;

[0099] 6c. For the feature points remaining after step 6b, if the distance from one endpoint to the other two endpoints is less than 4 pixels, then the endpoint is considered to be a pseudo feature point and is deleted.

[0100] like Fig. 9 Shown is an example of fingerprint feature points extracted according to the method of the present invention.

[0101] Example 2

[0102] The process of this method is as follows Figure 1 As shown:

[0103] 1. Pre-integrate the input fingerprint image I(x, y) to obtain the pre-integration graph G(x, y). Each element value G(x, y) in the pre-integration graph is as follows: Figure 2 The figure shows the sum of the upper left pixel values ​​of the pixel point (x, y), calculated according to the following formula:

[0104] G(x, y) = ∑ x′≤x，y′≤y I(x′, y′)

[0105] Where I(x′, y′) is the grayscale value of the pixel of the input fingerprint image (x′, y′).

[0106] 2. Estimate the local ridge direction of each pixel of the fingerprint image based on the gradient algorithm, including the following steps:

[0107] 2a. Calculate the gradient of the fingerprint image I(x, y) in and are the gradients in the x and y directions of pixel I(x, y), respectively, where:

[0108]

[0109] 2b. Calculate the local ridge direction based on the phase angle of the gradient: the direction perpendicular to the gradient direction is the direction of the ridge:

[0110]

[0111] like Figure 3 As shown, the left side is the input fingerprint image, and the right side is the direction estimation result.

[0112] 3. Binarize the input image I(x, y) according to the ridge direction to obtain a binary image B(x, y), where the value of B(x, y) is 1. The binarization steps are as follows:

[0113] 3a. For each pixel I(x, y) in the fingerprint image, take 9 pixels along the ridge line with the current pixel as the center and calculate the average value m l (x, y);

[0114] 3b. If Figure 4 As shown in the figure, the sum of the pixel values ​​in any rectangle in the gray area of ​​the input image I(x, y) is obtained by performing a few additions and subtractions based on the pre-integration matrix diagram:

[0115] A=G(x2,y2)-G(x1,y2)-G(x2,y1)+G(x1,y1)

[0116] The average value m of the face in the 9×9 neighborhood centered on the current pixel s (x, y), where the average value of the surface is m s (x, y) is obtained from the pre-integrated graph through several additions and subtractions and one division:

[0117]

[0118] Where G(x, y) is the value at coordinate (x, y) in the pre-integration graph.

[0119] 3c. Line mean m l The average gray value of (x, y) is less than the average gray value of the surface m s (x, y), then set B(x, y) to 1, that is, it is a ridge line, otherwise set B(x, y) to 0, that is, it is a valley line, such as Figure 5 As shown, the left side is the input fingerprint image, and the right side is the binary image after binarization through the above steps.

[0120] 4. Thin the binary image. The basic steps are as follows: for each pixel B(x, y) in the binary image, if B(x, y) = 1 and the following three conditions are met, set B(x, y) to 0.

[0121] 4a. C(p) = 1, indicating that the number of times the pixel p's 8-neighborhood changes from 0 to 1 in the clockwise direction is 1;

[0122] 4b.2≤min[N1(p),N2(p)]≤3, where

[0123] N1(p)=(p1∨p2)+(p3∨p4)+(p5∨p6)+(p7∨p8),

[0124] N2(p)=(p2∨p3)+(p4∨p5)+(p6∨p7)+(p8∨p1)

[0125] (4c)

[0126] Among them, Figure 6 As shown, p is the current pixel, and p1, ..., p8 are 8 fields of p in clockwise order starting from the upper left corner.

[0127] Repeat the process in step 4 for the binary image B(x, y) until no pixels are set to 0. Figure 7 The figure shows the result of thinning the binary image using the above method.

[0128] 5. Fingerprint feature point extraction, such as Figure 8 The fingerprint feature points shown include endpoints and bifurcation points. For each pixel point p in the refined image, p = 1 and: H(p) = 2, then p is an endpoint; H(p) = 6, then p is a bifurcation point, where:

[0129] Where p9 = p1

[0130] As in step 4, Figure 6 As shown, p is the current pixel, p1, ..., p8 are the 8 fields of p in clockwise order starting from the upper left corner.

[0131] 6. Delete pseudo feature points. Affected by noise and fingerprint cracks, many pseudo feature points will appear in the fingerprint image, which is particularly obvious at the edge of the fingerprint area. The feature points detected initially are screened, and the pseudo feature points are removed and the true feature points are retained. Pseudo feature points are deleted according to the following rules:

[0132] 6a. For each feature point extracted by the method in step 5, if it is close to the edge of the fingerprint area, that is, if any feature point appears in a non-fingerprint area within a radius of 6 pixels, the feature point is a pseudo feature point and is deleted;

[0133] 6b. For the feature points remaining after step 6a, if any two endpoints are in opposite directions and the distance is less than 5 pixels, then these two feature points are pseudo feature points and shall be deleted;

[0134] 6c. For the feature points remaining after step 6b, if the distance from one endpoint to the other two endpoints is less than 4 pixels, then the endpoint is considered to be a pseudo feature point and is deleted.

[0135] like Fig. 9 The figure shows an example of fingerprint feature points extracted according to the method of the present invention.

[0136] The present invention can quickly and accurately extract fingerprint feature points, and the most notable feature of the present invention is that it is simple and fast and does not require repeated cumulative calculations.

[0137] The method for fast binarization and feature extraction of fingerprint images based on pre-integrated graph and gradient direction described in the present invention is not limited to the description in the specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the scope of the claims of the present invention.

[0138] It should be noted that, in the claims, any reference signs placed between parentheses shall not be constructed as limiting the claims.The word "comprising" does not exclude the presence of elements or steps not listed in a claim.

[0139] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic creative concepts. The appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0140] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A fast fingerprint image binarization and feature extraction method, characterized in that: The steps include: The input fingerprint image I(x, y) is pre-integrated to obtain a pre-integrated graph G(x, y), where each element value G(x, y) in the pre-integrated graph is the sum of all pixel values ​​at the upper left corner of the pixel (x, y); The local ridge direction of each pixel of the fingerprint image is estimated based on the gradient algorithm; The input image I(x, y) is quickly binarized according to the ridge direction and the pre-integrated image to obtain a binary image B(x, y), where the value of B(x, y) is 0 or 1; Thinning the binary image B(x, y); Perform fingerprint feature point detection on the refined image; After deleting the pseudo feature points, the fingerprint feature file is obtained.

2. A fast fingerprint image binarization and feature extraction method according to claim 1, characterized in that: The fingerprint image binarization and feature extraction method includes a pre-integration method, which is as follows: each element value G (x, y) in the pre-integration graph is calculated according to the following formula: G(x,y)=∑ x′≤x，y′≤y I(x′,y′) Where I(x′, y′) is the grayscale value of the pixel of the input fingerprint image (x′, y′).

3. The method for rapid fingerprint image binarization and feature extraction according to claim 1, characterized in that: The fingerprint image binarization and feature extraction method includes a gradient-based ridge direction estimation method, and the gradient-based ridge direction estimation method includes the following steps: Calculate the gradient of the fingerprint image I(x, y) in and are the gradients in the x and y directions of pixel I(x, y), respectively, where: Calculate the local ridge direction based on the phase angle of the gradient; The direction perpendicular to the gradient is the direction of the ridge:

4. The method for rapid fingerprint image binarization and feature extraction according to claim 1, characterized in that: The fingerprint image binarization and feature extraction method includes a fingerprint image fast binarization method, and the fingerprint image fast binarization method includes the following steps: The steps of fingerprint image binarization are as follows: For each pixel I(x, y) in the fingerprint image, take a preset number of 2l+1 pixels along the ridge line with the current pixel as the center and calculate the average value m l (x, y); The average value m of the surface in the (2m+1)×(2n+1) neighborhood centered on the current pixel s (x, y), where the average value m s (x, y) is obtained from the pre-integrated graph through several additions and subtractions and one division: Where G(x, y) is the value at the coordinate (x, y) in the pre-integration graph; Line mean m l The average gray value of (x, y) is less than the average gray value of the surface m s (x, y), then set B(x, y) to 1, which means it is a ridge line, and set B(x, y) to 0, which means it is a valley line.

5. The method for rapid fingerprint image binarization and feature extraction according to claim 1, characterized in that: The fingerprint image binarization and feature extraction method includes a binarization image thinning method, and the binarization image thinning method includes the following steps: For each pixel B(x, y) in the binary image, if B(x, y) = 1 and the following three conditions are met, then B(x, y) is set to 0; Condition (1) C(p) = 1, indicating that the number of times the pixel p's 8-neighborhood changes from 0 to 1 in the clockwise direction is 1; Condition (2) 2≤min[N1(p), N2(p)]≤3, where N1(p)=(p1∨p2)+(p3∨p4)+(p5∨p6)+(p7∨p8), N2(p)=(p2∨p3)+(p4∨p5)+(p6∨p7)+(p8∨p1) Condition (3) Where p is the current pixel, p1, ..., p8 are the 8 fields of p in clockwise order starting from the upper left corner; The binary image B(x, y) is repeatedly processed according to the method described in step (4) until no pixels are set to 0.

6. A fast fingerprint image binarization and feature extraction method according to claim 1, characterized in that: The fingerprint image binarization and feature extraction method includes a fingerprint feature extraction method performed on a binarized image, and the fingerprint feature extraction method performed on a binarized image includes the following steps: For each pixel p in the refined image, if p = 1 and: H(p) = 2, then p is an endpoint; H(p) = 6, then p is a bifurcation point, where: Where p9 = p1 p is the current pixel, p1, ..., p8 are the 8 fields of p in clockwise order starting from the upper left corner.

7. A fast fingerprint image binarization and feature extraction method according to claim 1, characterized in that: The fingerprint image binarization and feature extraction method includes a pseudo feature point deletion method, and the pseudo feature point deletion method includes the following steps: For each candidate feature point, if it is close to the edge of the fingerprint area, the feature point is a pseudo feature point and is deleted; For any two endpoints of all candidate feature points, if their directions are opposite and their distance is less than the preset value, these two feature points are pseudo feature points and will be deleted. Among all the candidate feature points, if the distance from one endpoint to the other two endpoints is less than a preset value, the endpoint is considered to be a pseudo feature point and is deleted.

8. The method for rapid fingerprint image binarization and feature extraction according to claim 1, characterized in that: The fingerprint image binarization and feature extraction method includes a pre-integration image optimization method, and the pre-integration image optimization method includes the following steps: Pre-integration image size adjustment: adjust the size of the pre-integration image according to the resolution and size of the input fingerprint image; Block processing: Divide the pre-integrated image into multiple sub-blocks, and calculate each sub-block independently; Boundary processing: In the boundary area of ​​the pre-integration image, mirror filling or constant filling is used.

9. The method for rapid fingerprint image binarization and feature extraction according to claim 1, characterized in that: The fingerprint image binarization and feature extraction method includes a gradient algorithm optimization method, and the gradient algorithm optimization method includes the following steps: Multi-scale gradient calculation: Calculate the gradient of the fingerprint image at different scales to adapt to fingerprint images with different texture densities; Gradient direction smoothing: Smooth the gradient direction using Gaussian filtering or mean filtering; Directional consistency correction: In a local area, the direction of the current pixel is corrected according to the gradient direction of adjacent pixels.

Images