A method of fingerprint recognition

Abstract

The present application relates to a kind of fingerprint identification methods, residual fingerprint or spot fingerprint in fingerprint image, press deeper \ wet finger fingerprint, normal fingerprint identification can be realized, can improve identification accuracy, the method is based on fingerprint instrument system implementation, fingerprint instrument system includes fingerprint collection module, data processor, prompt module, the steps of processing, identifying to fingerprint image include: obtaining fingerprint image;Fingerprint image is preprocessed;The preprocessed image is blocked, obtains image block, calculates the gray mean and gray variance of each image block;Determine whether the fingerprint image is abnormal image or normal image;Determine whether the deep gray region in fingerprint image is pressed deeper \ wet finger fingerprint or spot map;The foreground region in fingerprint image is invalid fingerprint or normal fingerprint, if it is normal fingerprint or press deeper \ wet finger fingerprint, then through prompt module sends prompt, program stops, if it is invalid fingerprint, then prompt module does not send prompt.

Description

Technical Field

[0001] This invention relates to the field of fingerprint recognition technology, and more particularly to a fingerprint recognition method. Background Technology

[0002] A fingerprint scanner is an electronic device that uses the characteristics of a finger's fingerprint for identification. It typically consists of a lens group, a light source, a prism, an optical sensor, and a data processor. Currently, to save costs, fingerprint scanners generally do not determine whether a finger is pressed by using sensors (such as pressure-sensitive resistors or temperature-sensitive resistors). Instead, they continuously acquire images and use fingerprint detection algorithms to determine this. The fingerprint detection algorithm first processes the fingerprint image using an image processing algorithm, and then determines whether the fingerprint image is normal or abnormal based on the processing results. A normal image indicates a normal fingerprint, while an abnormal image indicates a residual fingerprint, a light spot, or no fingerprint. If a normal fingerprint is detected, a buzzer / speaker will be activated to notify the user that the fingerprint scanner is in normal working condition; otherwise, it indicates that the fingerprint scanner is in an abnormal working condition. However, in practical applications, when there are residual fingerprints on the lens group of the fingerprint scanner or when the fingerprint scanner is subjected to external environmental interference such as strong light, false triggering and continuous usage prompts may occur. The main reason for false triggering is the poor fingerprint recognition accuracy of the fingerprint detection algorithm.

[0003] Currently, commonly used fingerprint detection algorithms are mainly based on grayscale mean and grayscale variance. For example, existing technology provides a fingerprint recognition method based on optical fingerprint attendance machines, which can effectively distinguish between residual fingerprints and normal fingerprints in an image. However, in actual use, the fingerprint scanner is easily affected by the external environment when acquiring images, resulting in false triggering. For example, under sunlight, the acquired image may have abnormalities such as light spots. The existing fingerprint detection algorithm cannot accurately identify abnormal images containing light spots, causing the fingerprint scanner to be falsely triggered and continuously ringing, which seriously affects the customer experience. Summary of the Invention

[0004] To address the aforementioned technical problems in the existing technology, the present invention provides a fingerprint recognition method that can recognize residual fingerprints, light spots, fingerprints from deeply pressed / wet fingers, and normal fingerprints in fingerprint images, thereby reducing false triggering.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A fingerprint recognition method is provided, which is based on a fingerprint scanner system. The fingerprint scanner system includes a fingerprint acquisition module, a data processor, and a prompting module. The fingerprint acquisition module and the prompting module are both communicatively connected to the data processor. The fingerprint acquisition module includes a fingerprint sensor, which is used to acquire fingerprint images and send the acquired fingerprint images to the data processor. The data processor is used to process the fingerprint images and identify valid and invalid fingerprints in the fingerprint images. The invalid fingerprints are light spots, residual fingerprints, or no fingerprints. The valid fingerprints refer to normal fingerprints or fingerprints from deep pressing / wet fingers. The prompting module is used to issue prompts to the user based on the recognition results. The method is characterized in that the steps of processing and recognizing the fingerprint images include: S1, acquiring fingerprint images;

[0007] S2. Preprocess the fingerprint image to obtain a preprocessed image;

[0008] S3. Divide the preprocessed image into blocks to obtain image blocks, and calculate the mean gray level and gray level variance of each image block;

[0009] S4. Based on the grayscale variance and variance threshold of the image block, determine whether the fingerprint image is a normal image or an abnormal image. If it is an abnormal image, the program stops and the prompting module does not issue a prompt. If it is a normal image, proceed to step S5 and / or step S6.

[0010] S5. Determine whether the dark gray area in the fingerprint image is a deep press / wet finger fingerprint or a light spot. If it is a deep press / wet finger, issue a prompt through the prompt module and stop the program. If it is a light spot, the prompt module does not issue a prompt. The steps include: S51. Determine whether the image block is a dark gray area of ​​the fingerprint image. If it is, proceed to step S52. Otherwise, proceed to step S6.

[0011] S52. Determine whether the dark gray area in the fingerprint image is a light spot or a fingerprint of a deeply pressed / wet finger.

[0012] S6. Determine whether the foreground area in the fingerprint image is an invalid fingerprint or a normal fingerprint. If it is a normal fingerprint, issue a prompt through the prompting module and stop the program. If it is an invalid fingerprint, the prompting module will not issue a prompt.

[0013] Its further feature is that,

[0014] In step S2, the fingerprint image is preprocessed by sequentially using 3*3 mean filtering and grayscale stretching methods.

[0015] In step S3, the fingerprint image is divided into 8*8 blocks;

[0016] Furthermore, in step S4, the steps for obtaining valid blocks include: S41, setting a variance threshold;

[0017] S41. Determine whether the grayscale variance of the image block is greater than the variance threshold. If it is, it means that the image block belongs to the foreground region; otherwise, it means that the image block is the background region.

[0018] S42. Count the number of image blocks in the foreground region. If the number of image blocks is less than a certain percentage of the total number of image blocks after the entire fingerprint image is divided into blocks, then the fingerprint image is an abnormal image; otherwise, it is a normal image.

[0019] Furthermore, the variance threshold is 100;

[0020] Furthermore, the ratio is 0.2;

[0021] Furthermore, the total number of image blocks after dividing the entire fingerprint image into blocks is: BArea

[0022] =BWidth*BHeight, where BWidth is the number of row-oriented image blocks in a single fingerprint image, BWidth = imgWid / 8 = 32, imgWid is the width of the fingerprint image, imgWid = 256, BHeight is the number of column-oriented image blocks in a single fingerprint image, BHeight = imgHei / 8 = 36, imgHei is the height of the fingerprint image, imgHei = 288;

[0023] Furthermore, step S51 includes:

[0024] S511. Obtain the maximum value among the grayscale average values ​​of each image block. The maximum value is the grayscale value of the background area of ​​the image block.

[0025] S512. Traverse each image block. If the average gray value of an image block is less than the difference between the gray value of the background area of ​​the image block and a fixed value, then the image block is a dark gray area in the fingerprint image.

[0026] Furthermore, the difference is 80;

[0027] Furthermore, step S52 includes: S521, obtaining the connected components of the deep grayscale region, calculating the area of ​​each connected component, selecting the connected component with the largest area as the connected component of the deep grayscale region of the fingerprint image, and setting the grayscale value of the image block contained in the connected component to 0, and otherwise setting it to 1.

[0028] S522. Obtain the number of image blocks contained in each row of the largest connected component. Count the number of rows where the number of image blocks contained in each row of the largest connected component exceeds the width threshold (broadLineNum). Obtain the height of the largest connected component. If the larger value of the number of image blocks contained in the top row and bottom row of the largest connected component is greater than the width threshold, and the difference between the height of the largest connected component and the number of rows (broadLineNum) is less than the first threshold, then the fingerprint image is a light spot image. Otherwise, further judgment is needed on the fingerprint image. Proceed to step S523.

[0029] S523. If the difference between the height of the largest connected region and the number of rows broadLineNum is greater than or equal to the first threshold, and the height of the largest connected region is greater than 2*BHeight / 3, and the number of image blocks in the largest connected region exceeds 65% of the total number of image blocks in the entire fingerprint image, then the dark gray area in the fingerprint image is a deep press / wet finger fingerprint. Otherwise, further judgment is needed on the fingerprint in the fingerprint image, and proceed to step S6.

[0030] Furthermore, in step S6, the step of determining whether the foreground region in the fingerprint image is a normal fingerprint or an invalid fingerprint includes: S61, using a 3*3 filter to filter out isolated points in the valid fingerprint and obtain a first-filtered image;

[0031] S62. Delete the narrow regions in the primary filter image, including column-oriented narrow regions and row-oriented narrow regions, and obtain the secondary filter image;

[0032] S63. Erode the secondary filtered image to obtain the eroded image;

[0033] S64. Obtain the number of image blocks with a grayscale value of 0 in the erosion map;

[0034] S65. Determine whether the number of image blocks is less than a certain proportion of the total number of image blocks in the entire fingerprint image. If so, it indicates that the fingerprint in the fingerprint image is an invalid fingerprint; otherwise, proceed to step S66.

[0035] S66. Select the largest connected region with a value of 0 in the erosion map and calculate the number of image blocks in the largest connected region. If the number of image blocks in the largest connected region is less than a certain proportion of the total number of image blocks in the entire fingerprint image, then there is no valid fingerprint in the fingerprint image. Otherwise, it indicates that the foreground region in the fingerprint image is a normal fingerprint.

[0036] The above-described structure of this invention can achieve the following beneficial effects: In the fingerprint recognition method of this application, the fingerprint image is first recognized based on grayscale variance to determine whether the fingerprint image is an abnormal image or a normal image. Then, the fingerprint image is recognized a second time. The second recognition includes at least two methods: The first method is to accurately identify light spots or deep-pressed / wet fingerprints in the fingerprint image through step S5, thereby avoiding the problem of fingerprint scanner false triggering caused by the inaccurate recognition of light spots or deep-pressed / wet fingerprints; the second method is to distinguish between normal fingerprints and invalid fingerprints (invalid fingerprints are light spots, residual fingerprints, or no fingerprints) in the fingerprint image through step S6, thereby avoiding the problem of fingerprint scanner false triggering caused by the inaccurate recognition of invalid or normal fingerprints. The two recognition methods significantly reduce the false trigger rate. Attached Figure Description

[0037] Figure 1 This is a flowchart of the fingerprint recognition method of the present invention;

[0038] Figure 2 This is a grayscale image of a normal fingerprint.

[0039] Figure 3 A grayscale image showing the residual fingerprint / light spot under sunlight.

[0040] Figure 4 This is a grayscale image of a fingerprint remaining under conditions without sunlight.

[0041] Figure 5 This is a grayscale representation of the light spot image;

[0042] Figure 6 Grayscale image of a deep or wet fingerprint;

[0043] Figure 7 This is a diagram showing the filtered image obtained by applying a 3*3 mean filter to the fingerprint recognition method of the present invention.

[0044] Figure 8 The fingerprint recognition method of this invention uses a grayscale stretching method to... Figure 7 The image shows the result of grayscale transformation on the filtered image;

[0045] Figure 9 In the figures 9a, 9b, 9c, and 9d, the results of processing the grayscale transformed image after grayscale transformation using step S4 of the fingerprint recognition method of the present invention are: normal fingerprint, residual fingerprint, light spot, and fingerprint of a deeply pressed / wet finger.

[0046] Figure 10 This is an image showing the effect of a deeply pressed / wet finger fingerprint in the fingerprint image obtained after processing by steps S511 to S516 of the fingerprint recognition method of the present invention.

[0047] Figure 11 The image shows the effect of the light spot after processing by steps S521 to S523 of the fingerprint recognition method of the present invention.

[0048] Figure 12 Figures 12a, 12b, and 12c are effect diagrams of normal fingerprints, residual fingerprints, and light spots after removing isolated points using step S61 of the fingerprint recognition method of the present invention.

[0049] Figure 13 In the image, 13a, 13b, and 13c represent the fingerprints removed in step S621 of the fingerprint recognition method of the present invention. Figure 12 Images of normal fingerprints, residual fingerprints, and narrow areas in light spots, a, 12b, and 12c;

[0050] Figure 14 In the image, 14a, 14b, and 14c represent the fingerprints removed in step S622 of the fingerprint recognition method of the present invention. Figure 13 Images of normal fingerprints, residual fingerprints, and narrow areas in light spots in fingerprints a, 13b, and 13c;

[0051] Figure 15 In the diagram, 15a, 15b, and 15c represent the fingerprint recognition method of the present invention, step S63. Figure 14 The images shown depict a normal fingerprint, a residual fingerprint, and the effect of light spot etching. Detailed Implementation

[0052] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0053] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification, claims and accompanying drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product or device.

[0054] Currently, there is a lack of fingerprint detection algorithms capable of accurately recognizing light spot images under sunlight. Furthermore, in practical use, existing fingerprint scanners suffer from another type of false triggering: users often press their fingerprints at varying depths or with wet fingers (i.e., fingers with sweat or water). Deeper pressing or wet fingers can easily cause fingerprint adhesion, resulting in a smaller variance in the acquired fingerprint image (although the grayscale variance of fingerprints from deeper pressing or wet fingers is greater than that of residual fingerprints without sunlight). Existing fingerprint detection algorithms cannot accurately recognize inconsistent pressing depths or wet fingers, leading to the fingerprint scanner failing to trigger (i.e., issue a prompt or indication) when it should. Additionally, under sunlight, the grayscale variance of the light spot in the fingerprint image is also smaller (greater than that of residual fingerprints without sunlight), making it difficult to accurately distinguish between fingerprints from deeper pressing or wet fingers and the light spot, causing the fingerprint scanner to trigger falsely.

[0055] To address the technical problem of poor recognition accuracy of existing fingerprint detection algorithms for light spot images under sunlight and fingerprints from deeply pressed / wet fingers, the following is a specific embodiment of a fingerprint recognition method. This method is based on a fingerprint scanner system, which includes a fingerprint acquisition module, a data processor, and a prompting module. Both the fingerprint acquisition module and the prompting module are communicatively connected to the data processor. The fingerprint acquisition module includes a fingerprint sensor, which acquires fingerprint images and sends them to the data processor. The data processor processes the fingerprint images and identifies valid and invalid fingerprints. Invalid fingerprints are light spots or residual fingerprints, while valid fingerprints are normal fingerprints or fingerprints from deeply pressed / wet fingers. Light spot images refer to images with light spots acquired by the fingerprint sensor under sunlight. The prompting module provides prompts to the user based on the recognition results. In this embodiment, the prompting module is a buzzer or a speaker. When the fingerprint in the image acquired by the fingerprint sensor is a valid fingerprint, the buzzer sounds, indicating that the fingerprint scanner has been triggered normally. Conversely, when the fingerprint in the image acquired by the fingerprint sensor is an invalid fingerprint or no fingerprint information has been acquired, the buzzer does not sound.

[0056] The specific steps for processing and recognizing fingerprint images based on the above fingerprint scanner system include:

[0057] S1. Obtain fingerprint images through a data processor. The fingerprint images include normal fingerprint images (see...). Figure 2 Fingerprint images under sunlight (reference) Figure 3 Residual fingerprint images under no sunlight (reference) Figure 4 ), light spot image (reference) Figure 5 Deeper pressed fingerprint / wet finger fingerprint image (reference) Figure 6 ).

[0058] After dividing the image into blocks, the gray-level variance of each block is calculated sequentially. The gray-level variance effectively determines the gray-level variations within a block, thus allowing us to determine whether a block is a foreground or background region. Figures 2-6 The differences show that while residual fingerprint images, light spot images, and images of deeply pressed / wet fingers contain image patches with drastic gray-level changes in the foreground region, and there are quite a few of these patches, their continuity is poor. In contrast, the foreground region of a normal fingerprint image exhibits better continuity. Therefore, the presence of anomalies in a fingerprint image can be determined by examining the gray-level variance and the continuity (i.e., connected components) of the foreground region. Figure 4 It can be seen that in the absence of sunlight, the foreground (i.e., the residual fingerprint) in the residual fingerprint image has low contrast with the background, low variance, and a small number of effective image patches. However, from... Figure 3 It can be seen that if the fingerprint sensor's acquisition end (i.e., the lens group) in a fingerprint scanner is exposed to strong sunlight, the fingerprint image will contain many small black dots and bright dots with abrupt changes. These small black dots and bright dots increase the grayscale variance of the fingerprint image and increase the number of effective image blocks. This can lead to errors in the fingerprint recognition detection algorithm, causing the fingerprint scanner to falsely trigger an alarm sound. Figure 5 As can be seen, the foreground and background in the light spot image are quite different, and the grayscale variance is large. This can easily lead to fingerprint recognition detection algorithm errors and cause the fingerprint scanner to be triggered falsely.

[0059] S2. Preprocess the fingerprint image to obtain a preprocessed image. The specific steps of preprocessing include: S21. Filter the fingerprint image using a 3*3 mean filtering method. The filtered image is shown below. Figure 7 As shown, from Figure 7 It can be seen that small black dots, small bright dots and other abrupt changes in the residual fingerprint image under sunlight are effectively filtered out. This helps to reduce the impact of noise such as abrupt changes on gray-level variance and connected components, thereby improving the accuracy of fingerprint recognition.

[0060] S22. Perform grayscale stretching on the filtered image: Calculate the grayscale histogram of the filtered image, find the minimum grayscale value minGray and the maximum grayscale value maxGray, and then perform grayscale transformation, that is, stretch the grayscale range of the fingerprint image from [minGray, maxGray] to [0, 255]. The stretched image of a normal fingerprint after grayscale stretching is used as a reference. Figure 8 ,from Figure 8It can be seen that grayscale stretching improves the problem of low contrast between the foreground region (i.e., fingerprint) and the background region in fingerprint images. In other words, grayscale stretching enhances the contrast between the background and the foreground in fingerprint images, which is beneficial to improving the accurate recognition of fingerprints in the foreground region.

[0061] S3. Divide the preprocessed image into blocks to obtain image blocks. Calculate the grayscale mean and grayscale variance of each image block. The fingerprint image has a width (imgWid) of 256 and a height (imgHei) of 288. After dividing the fingerprint image into 8*8 non-overlapping blocks, the number of row-oriented image blocks is BWidth = imgWid / 8 = 32, and the number of column-oriented image blocks is BHeight = imgHei / 8 = 36. The total number of image blocks is BArea = BWidth * BHeight. Then, calculate the grayscale mean and grayscale variance of each image block and store them in the MeanTab and VarTab arrays, respectively. Existing fingerprint detection algorithms mainly rely on the global grayscale mean and grayscale variance of the fingerprint image for fingerprint recognition. However, this application first divides the fingerprint image into blocks and then extracts and recognizes the foreground region fingerprint by statistically analyzing the grayscale variance of each image block. This method yields more accurate foreground region fingerprints, which helps improve fingerprint recognition accuracy.

[0062] S4. Based on the grayscale variance and variance threshold of the image block, determine whether the image block is a foreground region or a background region. The foreground region is taken as a valid block. The valid blocks are counted and put into the array VaildResult. Based on the number of valid blocks, determine whether the fingerprint image is an abnormal image or a normal image. If it is an abnormal image, the program stops and the prompting module does not issue a prompt. If it is a normal image, proceed to step S5. Step S4 realizes the first fingerprint recognition in the fingerprint image.

[0063] Step S4 specifically includes: S41, setting a variance threshold, which is set to 100 in this embodiment;

[0064] S41. Traverse each image patch and determine if the grayscale variance of the image patch is greater than 100. If it is, the image patch belongs to the foreground region; otherwise, it belongs to the background region. Record the determination result in the VaildResult array: if the image patch at coordinate (i) is determined to be a foreground region, then VaildResult[i] = 0 (i.e., the grayscale value at coordinate (i) is set to 0); otherwise, it is set to 1. (Refer to...) Figure 9 ;

[0065] S42. Count the number of image blocks belonging to the foreground region, VaildSum. If the number of image blocks, VaildSum, is less than a certain proportion of the total number of image blocks, BArea, in a single fingerprint image, i.e., VaildSum < 0.2 * BArea, then the fingerprint image is an abnormal image, the program stops, and the indicator module does not issue a prompt. Otherwise, it is a normal image, realizing the first recognition of the fingerprint image, and proceeding to step S5 to perform a second recognition of the fingerprint image.

[0066] The process involves determining whether a dark gray area in the fingerprint image represents a deep press / wet finger fingerprint or a light spot. If the dark gray area in the fingerprint image represents a deep press / wet finger, a prompt is issued through the prompting module, and the program stops. If the dark gray area in the fingerprint image represents a light spot, the prompting module does not issue a prompt. The steps include: S51, determining whether the image block is a dark gray area of ​​the fingerprint image. If so, proceed to step S52; otherwise, proceed to step S6.

[0067] S51 includes: S511, obtaining the maximum value of the gray mean of each image block, that is, the maximum value in the MeanTab array, the maximum value is the gray value corMean of the background area of ​​the image block;

[0068] S512. Traverse each image block. If the grayscale mean value MeanTab[i] of the image block is less than the difference between the grayscale value corMean of the background area of ​​the image block and a fixed value (the fixed value is 80 in this embodiment), that is, MeanTab[i] < corMean-80, where i = 0, ..., BArea-1, then the image block is a dark grayscale region in the fingerprint image. Record the judgment result in the pDarkRegion array: if the image block at coordinate (i) is judged to be a dark grayscale region, then pDarkRegion[i] = 0 (that is, the grayscale at coordinate (i) is set to 0), otherwise it is set to 1.

[0069] The steps for determining whether a dark gray area in a fingerprint image is a light spot or a fingerprint from a deeply pressed / wet finger include: S521, To prevent misjudgment due to multiple dark gray areas in the fingerprint image, firstly, obtain the connected regions of the dark gray areas: connected regions with a value of 0 in the pDarkRegion array, calculate the area of ​​each connected region, and select the connected region with the largest area, maxDarkRegion, as the connected region of the dark gray area in this fingerprint image, refer to... Figure 10 , Figure 11 ;

[0070] S522. Obtain the number of image blocks lDarkNum in each row of the largest connected component. Count the number of rows where the number of image blocks lDarkNum exceeds the row-wise image block count threshold (i.e., BWidth-3), and obtain the height darkH of the largest connected component. If the larger value EdgelDarkLine of the lDarkNum in the top row and bottom row of the largest connected component is greater than the row-wise image block count threshold (the larger value EdgelDarkLine refers to comparing the lDarkNum of the top row and the same lDarkNum in the bottom row of the dark grayscale connected component and taking the largest value), and the difference between the height darkH (darkH is the total number of rows in the largest connected component) and the number of rows broadLineNum is less than the first threshold (i.e., darkH-broadLineNum<5), then the fingerprint image is a light spot pattern. See [link to relevant documentation]. Figure 11 At this time, the prompting module does not emit a prompting sound, thereby realizing the light spot recognition and avoiding the fingerprint scanner from being falsely triggered due to incorrect light spot recognition. Otherwise, the fingerprint image needs to be further judged and the process proceeds to step S523.

[0071] S523. If the difference between the height of the largest connected component (darkH) and the number of rows (broadLineNum) is greater than or equal to the first threshold (i.e., darkH - broadLineNum ≥ 5), and the height of the largest connected component is greater than 2 * BHeight / 3 (i.e., darkH > 2 * BHeight / 3), and the number of image blocks in the largest connected component exceeds 65% of the total number of image blocks (BArea), then the dark gray area in the fingerprint image is a fingerprint from a deeply pressed / wet finger. See [link / details]. Figure 10 This enables accurate recognition of fingerprints from deep presses or wet fingers in the fingerprint image. At this point, the prompt module is triggered and emits a prompt sound, and the program ends.

[0072] Based on the processing result of step S4, the fingerprint image is further judged in step S6 as follows: S6, determine whether the foreground area in the fingerprint image is a normal fingerprint or a residual fingerprint. If the fingerprint in the fingerprint image is a normal fingerprint, a prompt is issued through the prompting module. If the foreground area in the fingerprint image is a residual fingerprint or a light spot, the prompting module does not issue a prompt.

[0073] The steps for determining whether the foreground region in a fingerprint image is a normal fingerprint, a residual fingerprint, or a light spot include: S61, using a 3*3 filter to filter out isolated points in the fingerprint image and obtaining a first-pass filter image. The specific steps for filtering out isolated points include: S611, filtering out isolated points in the background region image blocks of the fingerprint image, including: traversing the VaildResult array; if VaildResult[i] equals 0, then calculating the number of image blocks with a value of 1 in its 3*3 neighborhood; if value≥5, then setting the gray value of VaildResult[i] to 1, that is, adjusting the image block at coordinate (i) to invalid.

[0074] S612. Filter out isolated points in the foreground region image blocks of the fingerprint image, including: traversing the VaildResult array. If VaildResult[i] equals 1, calculate the number of image blocks with a value of 1 in its 3*3 neighborhood. If value≤3, set VaildResult[i] to 0, that is, adjust the image block at coordinate (i) to be valid.

[0075] After processing steps S611 and S612, the number of messy points on the ValidResult array will be significantly reduced. Figure 12 ,from Figure 12 As can be seen, after step S61, the main area of ​​the normal fingerprint is largely preserved, while a large number of blank areas, thin strips, and small blocks appear in the middle area of ​​the residual fingerprint and light spot. In response to these situations, step S62 is used to further process the first-filtered image obtained in step S61.

[0076] S62. Delete the narrow regions in the first filter image, including column-oriented narrow regions and row-oriented narrow regions, and obtain the second filter image.

[0077] The steps for deleting narrow regions in a single filtered image include: S621, traversing the ValidResult array, counting the number of image patches tmpHei with a value of 0 in each column; if tmpHei is less than BHeight / 3, then setting all values ​​in that column of the ValidResult array to 1. See [link to relevant documentation]. Figure 13 ;

[0078] The steps for deleting narrow regions in a filtered image include: S622, traversing the VaildResult array, counting the number of image patches with a value of 0 in each row (tmpWid), and if tmpWid is less than BWidth / 3, setting all values ​​in that row of the VaildResult array to 1. See... Figure 14 ;

[0079] S623. Count the number tmpSum of image blocks with a value of 0 in the VaildResult array. If tmpSum < (BWidth * BHeight) / 5, it indicates that the current fingerprint image does not contain a fingerprint, and the program stops or returns.

[0080] After being processed through steps S61 and S62, from Figure 14 As can be seen from 14a, 14b, and 14c, most of the fingerprint information in the fingerprint image of a normal fingerprint can be retained with little change, while most of the effective information such as the light spot in the residual fingerprint and light spot image is filtered out with significant changes, and it is generally divided into multiple discontinuous regions.

[0081] S63. Erode the secondary filtered image to obtain an eroded image: Select a 3 * 3 window and perform an erosion operation. Refer to Figure 15 From Figure 15 As can be seen, after erosion with a 3 * 3 window, the area of the effective region of the normal fingerprint in the fingerprint image changes little, while the area of the effective region of the residual fingerprint and light spot decreases significantly, and multiple connected regions appear.

[0082] S64. Obtain the number of image blocks in the effective region, that is, the number VaildArea0 of image blocks with a value of 0 in the VaildResult array;

[0083] S65. Determine whether the number of image blocks VaildArea0 is less than BArea / 5. If so, it indicates that the fingerprint in this fingerprint image is an invalid fingerprint, thus realizing the identification of invalid fingerprints; if VaildArea0 ≥ BArea / 5, go to step S66;

[0084] S66. Select the largest connected domain with a value of 0 in the VaildResult array and calculate the number VaildAreal of image blocks in this largest connected domain. If VaildAreal < BArea / 5, there is no effective fingerprint in this fingerprint image; otherwise, it indicates that the fingerprint in this fingerprint image is a normal fingerprint, thus realizing the accurate identification of normal fingerprints. Through the above steps S61 - S66, the present application effectively distinguishes the light spot from the normal fingerprint in the fingerprint image.

[0085] In this embodiment, 1000 normal fingerprint images, 771 light spot images under different sunlight irradiation angles, and 137 residual fingerprint images are collected through the fingerprint sensor of the fingerprint instrument. After processing the fingerprint images using the above steps S1 - S6, the recognition accuracy rate of normal fingerprints and relatively deeply pressed fingerprints / wet fingerprints is 98%.

[0086] It is understood that the above detailed description of the present invention is for illustrative purposes only and is not intended to limit the technical solutions described in the embodiments of the present invention. Those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention to achieve the same technical effects; as long as the usage requirements are met, they are all within the protection scope of the present invention.

Claims

1. A fingerprint recognition method, the method being implemented based on a fingerprint scanner system, the fingerprint scanner system comprising a fingerprint acquisition module, a data processor, and a prompting module, wherein the fingerprint acquisition module and the prompting module are both communicatively connected to the data processor, the fingerprint acquisition module comprising a fingerprint sensor, the fingerprint sensor being used to acquire fingerprint images and send the acquired fingerprint images to the data processor, the data processor being used to process the fingerprint images and to identify valid and invalid fingerprints in the fingerprint images, wherein invalid fingerprints are light spots, residual fingerprints, or no fingerprints, and valid fingerprints refer to normal fingerprints or fingerprints from deep pressing / wet fingers, and the prompting module being used to issue prompts to the user based on the recognition results, characterized in that... The steps for processing and recognizing the fingerprint image include: S1, acquiring the fingerprint image; S2. Preprocess the fingerprint image to obtain a preprocessed image; S3. Divide the preprocessed image into blocks to obtain image blocks, and calculate the mean gray level and the variance gray level of each image block; S4. Based on the grayscale variance and variance threshold of the image block, determine whether the fingerprint image is a normal image or an abnormal image. If it is an abnormal image, the program stops and the prompting module does not issue a prompt. If it is a normal image, proceed to step S5 or step S6. S5. Determine whether the dark gray area in the fingerprint image is a deep press / wet finger fingerprint or a light spot. If it is a deep press / wet finger, issue a prompt through the prompt module and the program stops. If it is a light spot, the prompt module does not issue a prompt. The steps include: S51. Determine whether the image block is a dark gray area of ​​the fingerprint image. If it is, proceed to step S52. Otherwise, proceed to step S6. S52. Determine whether the dark gray area in the fingerprint image is a light spot or a fingerprint of a deeply pressed / wet finger. S6. Determine whether the foreground area in the fingerprint image is an invalid fingerprint or a normal fingerprint. If it is a normal fingerprint, issue a prompt through the prompting module and stop the program. If it is an invalid fingerprint, the prompting module will not issue a prompt. Step S52 includes: S521, obtaining the connected components of the deep grayscale region, calculating the area of ​​each connected component, selecting the connected component with the largest area as the connected component of the deep grayscale region of the fingerprint image, and setting the image blocks contained in the connected component to 0, otherwise setting them to 1; S522. Obtain the number of image blocks contained in each row of the connected component with the largest area. Count the number of rows where the number of image blocks contained in each row of the connected component with the largest area exceeds the width threshold (broadLineNum). Obtain the height of the connected component with the largest area. If the larger value of the number of image blocks contained in the top row and the bottom row of the connected component (EdgelDarkLine) is greater than the width threshold, and the difference between the height of the connected component with the largest area and the number of rows (broadLineNum) is less than the first threshold, then the fingerprint image is a light spot image. Otherwise, further judgment is needed on the fingerprint image. Proceed to step S523. S523. If the difference between the height of the largest connected component and the number of rows (broadLineNum) is greater than or equal to the first threshold, and the height of the largest connected component is greater than 2... If BHeight / 3, and the number of image blocks in the largest connected region exceeds 65% of the total number of image blocks in the entire fingerprint image, then the dark gray area in the fingerprint image is a deep press / wet finger fingerprint. Otherwise, further judgment is needed on the fingerprint in the fingerprint image, and proceed to step S6. In step S4, the step of determining whether the fingerprint image is a normal image or an abnormal image includes: S41, setting a variance threshold; S41. Determine whether the grayscale variance of the image block is greater than the variance threshold. If it is, the image in the image block belongs to the foreground region; otherwise, it belongs to the background region. S42. Count the number of image blocks in the foreground region. If the number of image blocks is less than a certain percentage of the total number of image blocks after dividing the entire fingerprint image into blocks, then the fingerprint image is an abnormal image; otherwise, it is a normal image.

2. The fingerprint recognition method according to claim 1, characterized in that, In step S2, 3 are used sequentially.

3. Mean filtering and grayscale stretching methods are used to preprocess fingerprint images.

3. The fingerprint recognition method according to claim 2, characterized in that, In step S3, the fingerprint image is divided into 8... 8 blocks.

4. The fingerprint recognition method according to claim 3, characterized in that, The variance threshold is 100, and the ratio is 0.

2.

5. The fingerprint recognition method according to claim 4, characterized in that, The total number of image blocks after dividing the entire fingerprint image is: BArea = BWidth BHeight, where BWidth is the number of row-oriented image blocks in a single fingerprint image, BWidth = imgWid / 8 = 32, imgWid is the width of the fingerprint image, imgWid = 256, BHeight is the number of column-oriented image blocks in a single fingerprint image, BHeight = imgHei / 8 = 36, imgHei is the height of the fingerprint image, imgHei = 288.

6. The fingerprint recognition method according to claim 5, characterized in that, Step S51 includes: S511. Obtain the maximum value among the grayscale average values ​​of each image block. The maximum value is the grayscale value of the background area of ​​the image block. S512. Traverse each image block. If the average gray value of an image block is less than the difference between the gray value of the background area of ​​the image block and a fixed value, then the image block is a dark gray area in the fingerprint image.

7. The fingerprint recognition method according to claim 6, characterized in that, The difference is 80.

8. The fingerprint recognition method according to claim 7, characterized in that, In step S6, the step of determining whether the foreground region in the fingerprint image is an invalid fingerprint or a normal fingerprint includes: S61, using 3 A filter of type 3 removes isolated points from valid fingerprints, resulting in a first-pass filter image. S62. Delete the narrow regions in the primary filter image, including column-oriented narrow regions and row-oriented narrow regions, and obtain the secondary filter image; S63. Erode the secondary filtered image to obtain the eroded image; S64. Obtain the number of image blocks with a value of 0 in the erosion map; S65. Determine whether the number of image blocks is less than a certain proportion of the total number of image blocks in the entire fingerprint image. If so, it indicates that the foreground area in the fingerprint image is an invalid fingerprint. Otherwise, proceed to step S66. S66. Select the largest connected region with a value of 0 in the erosion map and calculate the number of image blocks in the largest connected region. If the number of image blocks in the largest connected region is less than a certain proportion of the total number of image blocks in the entire fingerprint image, then there is no valid fingerprint in the fingerprint image. Otherwise, it indicates that the foreground region in the fingerprint image is a normal fingerprint.

Images