A comprehensive evaluation method and system for finger vein image quality

Abstract

The present invention discloses a comprehensive evaluation method and system for finger vein image quality. The method includes the following steps: obtaining multiple corrected finger vein images; filtering the corrected finger vein images using a nonlinear median filtering method; preliminarily screening the grayscale finger vein images using a grayscale contrast evaluation model; binarizing the grayscale finger vein images that pass the preliminary screening using a dynamic threshold segmentation method; obtaining a single-pixel binary finger vein image using a skeleton extraction method based on the binarized finger vein image and its connected domain marking characteristics; calculating a texture distribution saturation parameter; determining the number of finger vein burrs by determining the presence of finger vein burrs; determining the number of pseudo-vein textures by determining the pseudo-vein textures; constructing a finger vein image quality evaluation model based on the texture distribution saturation parameter, the number of finger vein burrs, and the number of pseudo-vein textures, and using the evaluation model to evaluate the quality of the finger vein image.

Description

Technical Field

[0001] The present invention relates to a comprehensive evaluation method for finger vein image quality and also to a corresponding comprehensive evaluation system for finger vein image quality, belonging to the technical field of biometric feature recognition. Background Art

[0002] With the continuous advancement of information technology, the security requirements for personal identity information are also increasing. Traditional identity authentication methods based on magnetic cards, ID cards, and passwords are prone to loss, theft, and duplication, making them unable to guarantee the security of personal identity information. In recent years, biometric recognition technology, as a personal identity authentication method that utilizes a person's inherent characteristics as an authentication identifier, has become increasingly widely used due to its excellent adaptability and security. Unlike first-generation biometric technologies such as fingerprint recognition or facial recognition, finger vein recognition technology extracts vein features from captured near-infrared images of finger veins and then performs pattern matching against a feature library to achieve identity authentication. Finger vein recognition technology, due to its inherent advantages such as liveness detection and internal features, is less susceptible to feature forgery. Compared with first-generation biometric technologies, it offers higher security, better stability, and more efficient identity authentication.

[0003] In existing technologies, most finger vein authentication system research is based on two-dimensional finger vein images. Due to the limited information contained in two-dimensional finger vein images, authentication systems struggle to maintain their high accuracy in larger user scenarios. Furthermore, during the finger vein image acquisition process, image quality issues such as unclear image boundaries often arise due to factors such as ambient lighting variations, differences in finger placement, and finger disturbances. This leads to high false rejection and false recognition rates, making it difficult for the authentication system to accurately identify the finger vein. Furthermore, because acquiring finger vein images requires additional near-infrared light sources and specialized image sensors, variations in the captured vein images caused by differences in finger vein acquisition devices deployed at different terminals can negatively impact authentication performance. Therefore, developing a comprehensive finger vein image quality assessment model that can rapidly assess the quality of captured finger vein images and screen out defective images is crucial for improving the accuracy and precision of existing finger vein recognition and authentication systems.

[0004] Chinese invention patent ZL 202010007188.8 discloses a method for quantitatively evaluating finger vein image quality. The method includes calculating a grayscale distribution index for finger vein images, a noise level index for finger vein images, a first-order gradient index for finger vein images, a second-order gradient index for finger vein images, constructing a mathematical model for quantitative evaluation of finger vein image quality and calculating model parameters. The constructed mathematical model for quantitative evaluation of finger vein image quality is then used to quantitatively evaluate image quality under different environments. Summary of the Invention

[0005] The primary technical problem to be solved by the present invention is to provide a comprehensive evaluation method for finger vein image quality.

[0006] Another technical problem to be solved by the present invention is to provide a comprehensive evaluation system for finger vein image quality.

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

[0008] According to a first aspect of an embodiment of the present invention, a comprehensive evaluation method for finger vein image quality is provided, comprising the following steps:

[0009] (1) obtaining multiple corrected finger vein images from a finger vein recognition system;

[0010] (2) filtering the corrected finger vein image using a nonlinear median filtering method to obtain a grayscale finger vein image;

[0011] (3) performing a preliminary screening of the grayscale finger vein image using a grayscale contrast evaluation model to obtain the grayscale finger vein image that passes the preliminary screening;

[0012] (4) using a dynamic threshold segmentation method to binarize the grayscale finger vein image that has passed the initial screening to obtain a binary finger vein image;

[0013] (5) Based on the binary finger vein image and its connected domain marking characteristics, a skeleton extraction method is used to obtain a single-pixel binary finger vein image;

[0014] (6) calculating the texture distribution saturation parameter of the single-pixel binarized finger vein image;

[0015] (7) determining the finger vein burrs, and obtaining the number of finger vein burrs in the single-pixel binary finger vein image;

[0016] (8) Obtaining the number of pseudo vein textures of the single-pixel binary finger vein image by determining the pseudo vein texture;

[0017] (9) A finger vein image quality assessment model is constructed based on the texture distribution saturation parameter, the number of finger vein burrs, and the number of pseudo-vein textures, and the quality of the finger vein image is assessed using the assessment model.

[0018] Preferably, step (1) includes the following sub-steps:

[0019] (11) obtaining an original finger vein video stream in a time dimension from the finger vein recognition system;

[0020] (12) Calibrate the infrared camera parameters in the finger vein recognition system using a planar chessboard calibration method to obtain the camera's intrinsic parameters and distortion coefficients;

[0021] (13) Using the intrinsic parameters and the distortion coefficients, an image distortion correction method is used to correct the original finger vein video stream to obtain a corrected finger vein video stream and the multiple finger vein images contained therein.

[0022] Preferably, in step (2), the grayscale finger vein image g(x, y) satisfies the following formula:

[0023] g(x,y)=med{f(xm,yn)};m,n∈[-w1,+w1]

[0024] Where med represents the median; f(x, y) represents the finger vein image to be filtered; w1 is a positive integer representing the half-width of the sliding window; m and n are integers, and (xm, yn) represents the coordinates of the pixel points in the sliding window.

[0025] Preferably, step (3) includes the following sub-steps:

[0026] (31) calculating a grayscale histogram of the grayscale finger vein image to obtain an average grayscale value;

[0027] (32) Based on the differences in grayscale values ​​at different pixel coordinate positions, a mean square error algorithm is used to construct the grayscale contrast evaluation model, and the grayscale contrast of the grayscale finger vein image is calculated;

[0028] (33) comparing the grayscale contrast with a grayscale contrast threshold to preliminarily screen the quality of the grayscale finger vein image;

[0029] When the grayscale contrast is greater than or equal to the grayscale contrast threshold, the quality of the current grayscale finger vein image meets the initial screening requirement; otherwise, the current grayscale finger vein image fails the initial screening.

[0030] Preferably, in step (32), the grayscale contrast C rThe calculation satisfies the following formula:

[0031]

[0032] Among them, v i is the gray value of the i-th pixel in the image; v m is the average gray value; N total is the total number of pixels in the image.

[0033] Preferably, in step (6), the calculation of the texture distribution saturation parameter α satisfies the following formula:

[0034]

[0035] Among them, k M 、k N are all positive integers, representing the equal divisions along the horizontal and vertical directions of the resolution of the single-pixel binary finger vein image; T r k M *k N The total number of block areas with finger vein texture information in the block areas.

[0036] Preferably, step (7) includes the following sub-steps:

[0037] (71) Determine the bifurcation point of the finger vein texture;

[0038] (72) determining the endpoints of the finger vein texture and calculating the pixel length from the bifurcation point to the adjacent endpoints;

[0039] (73) Determine the finger vein burrs in the finger vein texture and calculate the number of the finger vein burrs.

[0040] Preferably, in step (9),

[0041] The output value E of the finger vein image quality assessment model satisfies the following formula:

[0042] E=f(α,β,γ)

[0043] Among them, E is the output value of the model, which is a parameter representing the quality of the finger vein image; α is the texture distribution saturation parameter; β is the number of finger vein burrs; and γ is the number of pseudo-vein textures.

[0044] According to a second aspect of an embodiment of the present invention, a comprehensive evaluation system for finger vein image quality is provided, the system comprising a processor and a memory, wherein the processor and the memory are coupled;

[0045] The memory is used to store computer programs;

[0046] The processor is configured to run the computer program stored in the memory to execute the comprehensive evaluation method for finger vein image quality as described above.

[0047] Compared with existing technologies, the comprehensive assessment method for finger vein image quality provided by the present invention uses a grayscale contrast assessment model to initially screen finger vein images of low quality. Furthermore, based on the characteristics of vein texture structure and its information entropy, a finger vein image quality assessment model is constructed based on texture distribution saturation parameters, the number of vein burrs, and the number of pseudo-vein textures. This technical solution performs a secondary comprehensive assessment and screening of finger vein images. This ensures the quality of finger vein images subsequently used in finger vein recognition systems and reduces misidentification caused by poor image quality due to ambient lighting, finger position disturbances, and other factors. Therefore, the comprehensive assessment method for finger vein image quality provided by the present invention has the beneficial effects of strong environmental adaptability, high anti-interference ability, high recognition accuracy, and good stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flowchart of a comprehensive evaluation method for finger vein image quality provided by the present invention;

[0049] Figure 2 is a schematic diagram of a binarized finger vein image in an embodiment of the present invention;

[0050] Figure 3 Schematic diagram of a finger vein texture that is a finger vein burr in an embodiment of the present invention;

[0051] Figure 4 This is a schematic diagram of a finger vein texture being a pseudo vein in an embodiment of the present invention;

[0052] Figure 5 This is a schematic structural diagram of a comprehensive evaluation system for finger vein image quality provided by the present invention. DETAILED DESCRIPTION

[0053] The technical content of the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] like Figure 1 As shown, an embodiment of the present invention provides a comprehensive evaluation method for finger vein image quality, which includes at least the following steps:

[0055] S1: Acquire multiple rectified finger vein images from a finger vein recognition system.

[0056] When a finger vein recognition system identifies a person's finger, it uses a monocular infrared camera to generate a raw finger vein video stream in the time dimension, which includes multiple frames of raw finger vein images with timestamps. The acquisition of multiple corrected finger vein images includes the following sub-steps:

[0057] S11: Obtaining the original finger vein video stream in the time dimension from the finger vein recognition system.

[0058] S12: Use the plane chessboard calibration method to calibrate the infrared camera parameters in the finger vein recognition system to obtain the camera's intrinsic parameters K in And distortion coefficient Dist.

[0059] S13: Using the internal parameter K in and distortion coefficient Dist, and the image distortion correction method is used to correct the original finger vein video stream to obtain the corrected finger vein video stream and the multiple finger vein images f(x, y) it contains; where (x, y) is the coordinate value of the image pixel.

[0060] S2: Using a nonlinear median filtering method to filter the corrected finger vein image to obtain a grayscale finger vein image.

[0061] The goal of filtering the corrected finger vein images is to reduce the noise generated during the acquisition process while preserving the texture structure of the finger veins. The filtering method involves selecting a fixed-size sliding window [2w1+1,2w1+1] centered at the target pixel in each finger vein image, where w1 is a positive integer representing the half-width of the sliding window. The grayscale values ​​of the pixels within this sliding window are sorted, and the median value is taken as the new grayscale value for the target pixel, thus forming a grayscale finger vein image.

[0062] Assuming that the finger vein image to be filtered is f(x,y), and the grayscale finger vein image after filtering is g(x,y), the expression of the grayscale finger vein image g(x,y) is:

[0063] g(x,y)=med{f(xm,yn)};m,n∈[-w1,+w1] (1)

[0064] Among them, med means taking the median; m and n are integers, and (xm,yn) represents the coordinates of the pixel points in the sliding window.

[0065] For example, when w1=1, the sliding window [2w1+1,2w1+1] is represented as a 3×3 pixel area. With the target pixel position (x0, y0) as the center, the grayscale values ​​of the 9 pixel positions including the center point can be obtained as shown in Table 1.

[0066] Table 1

[0067] <![CDATA[f(x0-1,y0+1)]]> <![CDATA[f(x0,y0+1)]]> <![CDATA[f(x0+1,y0+1)]]> <![CDATA[f(x0-1,y0)]]> <![CDATA[f(x0,y0)]]> <![CDATA[f(x0+1,y0)]]> <![CDATA[f(x0-1,y0-1)]]> <![CDATA[f(x0,y0-1)]]> <![CDATA[f(x0+1,y0-1)]]>

[0068] Formula (1) indicates that by sorting the grayscale values ​​of the above 9 pixels from large to small, the median of the grayscale values ​​is selected as the new grayscale value of the target pixel position (x0, y0), thereby obtaining the grayscale finger vein image g(x, y).

[0069] S3: Using the grayscale contrast evaluation model to preliminarily screen the grayscale finger vein image, and obtain a grayscale finger vein image that passes the preliminary screening.

[0070] Since the clarity of the vein texture structure directly determines the quality of the finger vein image, in order to prevent blurry and difficult-to-identify finger vein images during the recognition process, it is necessary to build a grayscale contrast evaluation model before performing finger vein image quality assessment to pre-eliminate low-quality finger vein images. The specific method of initial screening includes the following sub-steps:

[0071] S31: Calculate the grayscale histogram of the grayscale finger vein image to obtain the average grayscale value v m ;

[0072] S32: Based on the differences in grayscale values ​​at different pixel coordinate positions, a mean square error algorithm is used to construct a finger grayscale contrast evaluation model, and the grayscale contrast of the grayscale finger vein image is calculated using the grayscale contrast evaluation model. Grayscale contrast C r The calculation is as follows:

[0073]

[0074] Among them, v i is the grayscale value of the i-th pixel in the image; N total is the total number of pixels in the image.

[0075] S33: The grayscale contrast C of the grayscale finger vein image is r Compared with the grayscale contrast threshold τ, the quality of the grayscale finger vein image is preliminarily screened. r When the grayscale contrast is greater than or equal to the grayscale contrast threshold τ, the quality of the current grayscale finger vein image is considered to meet the initial screening requirements; otherwise, the grayscale finger vein image is considered to be unqualified in the initial screening and is marked as a defective finger vein image.

[0076] S4: A dynamic threshold segmentation method is used to perform binarization processing on the grayscale finger vein image that has passed the initial screening to obtain a binary finger vein image.

[0077] Binarization of grayscale finger vein images that pass the initial screening can be used to characterize their quality from a binary perspective. This is achieved by selecting a corresponding neighborhood window [2w²+1,2w²+1] for each pixel in the image, where w² is a positive integer representing the half-width of the neighborhood window. The binarization threshold T(x,y) for segmentation within this neighborhood window is determined by calculating the mean m(x,y) and variance σ(x,y) of the pixels within the neighborhood window.

[0078] The calculation of the binary segmentation threshold T(x,y) is as follows:

[0079] T(x,y)=m(x,y)+kσ(x,y) (3)

[0080] Wherein, k is a preset correction coefficient; (x, y) is the coordinate value of the image pixel.

[0081] By traversing different regions of the image through the above method, the binary processing of the overall grayscale finger vein image is completed, such as Figure 2 As shown, a binary finger vein image with coarser vein texture can be obtained.

[0082] S5: Based on the binary finger vein image and its connected domain marking characteristics, a skeleton extraction method is used to obtain a single-pixel binary finger vein image.

[0083] In order to achieve accurate characterization of the topological structure of finger vein features with more stable characteristics, the present invention adopts a skeleton extraction method based on the binary finger vein image and its binary connected domain marking characteristics to further obtain a finger vein texture map with a single-pixel width, that is, the single-pixel binary finger vein image to be evaluated.

[0084] In the quality evaluation process of single-pixel binary finger vein images, considering the imaging and subsequent algorithm matching characteristics of finger veins, the present invention proposes to construct an image quality assessment model based on texture distribution saturation, vein burr noise, and pseudo-vein texture from the perspective of finger vein texture distribution characteristics and its corresponding information entropy, and perform a secondary evaluation of vein image quality.

[0085] S6: Calculate the texture distribution saturation parameter of the single-pixel binarized finger vein image.

[0086] In terms of the saturation of the finger vein texture distribution, the texture information entropy is characterized based on the uniformity of the texture distribution of the finger vein image. The specific calculation method of the texture distribution saturation parameter is as follows.

[0087] Assume that the resolution of the single-pixel binary finger vein image to be evaluated is [M, N], where M and N are both positive integers, representing the horizontal and vertical resolutions respectively; k M and k NEqually divided (k M and k N is a positive integer), the whole image can be divided into k m *k n block areas, and by judging whether each block area has finger vein texture information, the total number of block areas with finger vein texture information T is obtained by statistics. r , and finally the texture distribution saturation parameter α is obtained, which is calculated as follows:

[0088]

[0089] S7: Obtaining the number of finger vein burrs in the single-pixel binary finger vein image by determining the finger vein burrs.

[0090] Regarding finger vein burr noise, the edges of binary vein images often have tiny venous burrs, which can adversely affect the matching and representation of the actual finger vein texture structure. Therefore, the present invention introduces the number of finger vein burrs as a parameter for quality evaluation. The specific implementation method includes the following sub-steps:

[0091] S71: Determine the bifurcation point of the finger vein texture;

[0092] Taking the finger vein texture end at the upper left corner of the single-pixel binary finger vein image as the starting point, the pixel value of each pixel in the 8 adjacent neighborhoods is traversed in sequence based on the connected component marking feature. If only one pixel value in the 8 neighborhoods is non-zero, the tracking continues along the non-zero point until the number of non-zero points in the 8 neighborhoods of the non-zero point P is greater than 1. In this case, the non-zero point P is considered to be the bifurcation point of the finger vein texture.

[0093] S72: Determine the endpoints of the finger vein texture and calculate the pixel length from the bifurcation point to the adjacent endpoint.

[0094] Re-track with the bifurcation point P as the starting point and continue tracking according to the traversal method of the connected domain marking feature mentioned above until there are no new texture pixels. It is considered that the tracking has reached the endpoint of the finger vein texture. By calculating the pixel length from the bifurcation point to the adjacent endpoint, the finger vein burr is judged.

[0095] S73: determining the finger vein burrs in the finger vein texture and counting the number of the finger vein burrs;

[0096] The method for judging finger vein burrs is to obtain the pixel length threshold L from the bifurcation point to the adjacent endpoint in the standard finger vein texture through statistical analysis. th In the finger vein texture of the current image, if the pixel length from the bifurcation point to the adjacent endpoint is less than the corresponding pixel length threshold L th , then the finger vein texture is judged to be a finger vein burr, such as Figure 3 As shown in the figure, the number of finger vein burrs β in the whole image can be obtained by judging and counting them one by one.

[0097] S8: Obtaining the number of pseudo vein textures of the single-pixel binary finger vein image by determining the pseudo vein texture.

[0098] In terms of pseudo-vein texture, based on the spatial continuity assumption of finger vein images, isolated pseudo-vein textures are judged and counted respectively.

[0099] The specific implementation method is similar to the above-mentioned method for judging finger vein burrs. By counting the pixel points with a pixel value of 1 in each texture feature connected area in the single-pixel binary finger vein image, the pixel length of the isolated finger vein texture is obtained; then the pixel length of the isolated finger vein texture is compared with the pixel length threshold L of the isolated finger vein texture. f The pseudo vein texture is judged by comparing the pixel length threshold L of the isolated finger vein texture. f It is preset based on the resolution of the acquired finger vein image.

[0100] If the pixel length of the isolated finger vein texture is less than the corresponding pixel length threshold L f , then the finger vein texture is judged to be a pseudo vein texture, such as Figure 4 By traversing all pixels of the image in sequence, the number of pseudo vein textures γ on the entire finger vein image can be obtained.

[0101] S9: A finger vein image quality assessment model is constructed based on the texture distribution saturation parameter α, the number of finger vein burrs β, and the number of pseudo vein textures γ, and the quality of the vein image is assessed using the assessment model.

[0102] The expression of the output value E of the evaluation model is:

[0103] E=f(α,β,γ) (5)

[0104] Among them, E is the output value of the evaluation model, which refers to the characterization parameter of vein image quality.

[0105] The evaluation model E calculates a corresponding characterization parameter E for finger vein image quality based on the saturation parameter of the finger vein texture distribution, the number of finger vein burrs, and the number of pseudo-vein textures. The quality of the finger vein image is assessed by comparing the characterization parameter E output by the evaluation model with a characterization parameter threshold E0. The characterization parameter threshold E0 is a preset characterization parameter threshold for finger vein image quality based on a statistical analysis of the quality of typical finger vein images.

[0106] When the characterization parameter E output by the evaluation model is greater than or equal to the characterization parameter threshold E0, the quality of the current finger vein image is determined to meet the requirements of subsequent recognition processing; otherwise, the current finger vein image is judged to be of low quality and is filtered out. By comprehensively evaluating the quality of multiple vein images through the evaluation model, the accuracy and stability of subsequent finger vein recognition system finger vein image recognition can be improved.

[0107] The above is a detailed description of a comprehensive evaluation method for finger vein image quality provided by the present invention. Based on the comprehensive evaluation method for finger vein image quality, an embodiment of the present invention further provides a comprehensive evaluation system for finger vein image quality, such as Figure 5 As shown, the comprehensive evaluation system includes one or more processors and a memory. The memory is coupled to the processor and is used to store one or more computer programs. When the one or more computer programs are executed by the one or more processors, the one or more processors implement the comprehensive evaluation method for finger vein image quality in the above-mentioned embodiment.

[0108] The processor is used to control the overall operation of the comprehensive finger vein image quality assessment system to complete all or part of the steps of the comprehensive finger vein image quality assessment method described above. The processor module can be a central processing unit (CPU), a graphics processing unit (GPU), a field programmable gate array (FPGA), an application-specific integrated circuit (ASIC), a digital signal processing (DSP) chip, etc. The memory is used to store various types of data to support operations on the comprehensive finger vein image quality assessment system. This data can include, for example, instructions for any application or method used to operate the comprehensive finger vein image quality assessment system, as well as application-related data. The memory module can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, etc.

[0109] The above-mentioned comprehensive evaluation system for finger vein image quality can improve the accuracy and stability of finger vein image recognition by subsequent finger vein recognition systems by performing primary screening and secondary evaluation on multiple finger vein images provided by the finger vein recognition system, removing finger vein images with poor imaging quality and retaining finger vein images that are judged to be qualified by comprehensive evaluation.

[0110] Compared with existing technologies, the comprehensive assessment method for finger vein image quality provided by the present invention uses a grayscale contrast assessment model to initially screen finger vein images of low quality. Furthermore, based on the characteristics of vein texture structure and its information entropy, a finger vein image quality assessment model is constructed based on texture distribution saturation parameters, the number of vein burrs, and the number of pseudo-vein textures. This technical solution performs a secondary comprehensive assessment and screening of finger vein images. This ensures the quality of finger vein images subsequently used in finger vein recognition systems and reduces misidentification caused by poor image quality due to ambient lighting, finger position disturbances, and other factors. Therefore, the comprehensive assessment method for finger vein image quality provided by the present invention has the beneficial effects of strong environmental adaptability, high anti-interference ability, high recognition accuracy, and good stability.

[0111] It should be noted that, in the description of the present invention, “plurality” means two or more than two, unless otherwise clearly defined.

[0112] The above describes in detail the comprehensive finger vein image quality assessment method and system provided by the present invention. For those skilled in the art, any obvious modification to this invention without departing from its essence would constitute an infringement of the present invention's patent rights and would result in corresponding legal liability.

Claims

1. A comprehensive evaluation method for finger vein image quality, characterized by The steps include: (1) obtaining multiple corrected finger vein images from a finger vein recognition system; (2) filtering the corrected finger vein image using a nonlinear median filtering method to obtain a grayscale finger vein image; (3) performing a preliminary screening of the grayscale finger vein image using a grayscale contrast evaluation model to obtain the grayscale finger vein image that passes the preliminary screening; (4) using a dynamic threshold segmentation method to binarize the grayscale finger vein image that has passed the initial screening to obtain a binary finger vein image; (5) Based on the binary finger vein image and its connected domain marking characteristics, a skeleton extraction method is used to obtain a single-pixel binary finger vein image; (6) calculating the texture distribution saturation parameter of the single-pixel binarized finger vein image; (7) determining the finger vein burrs, and obtaining the number of finger vein burrs in the single-pixel binary finger vein image; (8) Obtaining the number of pseudo vein textures of the single-pixel binary finger vein image by determining the pseudo vein texture; (9) A finger vein image quality assessment model is constructed based on the texture distribution saturation parameter, the number of finger vein burrs, and the number of pseudo-vein textures, and the quality of the finger vein image is assessed using the assessment model.

2. The comprehensive evaluation method for finger vein image quality according to claim 1, characterized in that Step (1) includes the following sub-steps: (11) obtaining an original finger vein video stream in a time dimension from the finger vein recognition system; (12) Calibrate the infrared camera parameters in the finger vein recognition system using a planar chessboard calibration method to obtain the camera's intrinsic parameters and distortion coefficients; (13) Using the intrinsic parameters and the distortion coefficients, an image distortion correction method is used to correct the original finger vein video stream to obtain a corrected finger vein video stream and the multiple finger vein images contained therein.

3. The comprehensive evaluation method for finger vein image quality according to claim 1, characterized in that In step (2), the grayscale finger vein image g(x,y) satisfies the following formula: g(x,y)=med{f(xm,yn)};m,n∈[-w1,+w1] Where med represents the median; f(x, y) represents the finger vein image to be filtered; w1 is a positive integer representing the half-width of the sliding window; m and n are integers, and (xm, yn) represents the coordinates of the pixel points in the sliding window.

4. The comprehensive evaluation method for finger vein image quality according to claim 1, characterized in that Step (3) includes the following sub-steps: (31) calculating a grayscale histogram of the grayscale finger vein image to obtain an average grayscale value; (32) Based on the differences in grayscale values ​​at different pixel coordinate positions, a mean square error algorithm is used to construct the grayscale contrast evaluation model, and the grayscale contrast of the grayscale finger vein image is calculated; (33) comparing the grayscale contrast with a grayscale contrast threshold to preliminarily screen the quality of the grayscale finger vein image; When the grayscale contrast is greater than or equal to the grayscale contrast threshold, the quality of the current grayscale finger vein image meets the initial screening requirement; otherwise, the current grayscale finger vein image fails the initial screening.

5. The comprehensive evaluation method for finger vein image quality according to claim 4, characterized in that In step (32), the grayscale contrast C r The calculation satisfies the following formula: Among them, v i is the gray value of the i-th pixel in the image; v m is the average gray value; N total is the total number of pixels in the image.

6. The comprehensive evaluation method for finger vein image quality according to claim 1, characterized in that In step (6), the calculation of the texture distribution saturation parameter α satisfies the following formula: Among them, k M 、k N are all positive integers, representing the equal divisions along the horizontal and vertical directions of the resolution of the single-pixel binary finger vein image; T r k M *k N The total number of block areas with finger vein texture information in the block areas.

7. The comprehensive evaluation method for finger vein image quality according to claim 1, characterized in that Step (7) includes the following sub-steps: (71) Determine the bifurcation point of the finger vein texture; (72) determining the endpoints of the finger vein texture and calculating the pixel length from the bifurcation point to the adjacent endpoints; (73) Determine the finger vein burrs in the finger vein texture and calculate the number of the finger vein burrs.

8. The comprehensive evaluation method for finger vein image quality according to claim 1, characterized in that In step (9), the output value E of the finger vein image quality assessment model satisfies the following formula: E=f(α,β,γ) Among them, E is the output value of the model, which is a parameter representing the quality of the finger vein image; α is the texture distribution saturation parameter; β is the number of finger vein burrs; and γ is the number of pseudo-vein textures.

9. A comprehensive evaluation system for finger vein image quality, characterized in that The invention comprises a processor and a memory, wherein the processor and the memory are coupled; wherein the memory is used to store a computer program; and the processor is used to run the computer program stored in the memory to execute the comprehensive evaluation method for finger vein image quality according to any one of claims 1 to 8.

Images