Biological feature recognition method based on finger vein image layering

Abstract

The invention provides a biological feature recognition method based on finger vein image layering, and relates to the technical field of biological recognition, and the method comprises the steps: collecting a finger vein image of a to-be-recognized person, carrying out the preprocessing of the finger vein image, obtaining a first recognition image of finger vein layering, extracting a region of interest of a finger through an edge detection algorithm, and carrying out the recognition of the finger vein layering; a continuous edge curve is formed through double-threshold detection, after feature region segmentation is achieved, surface vein texture features are extracted, deep vein geometric features are extracted, then finger vein index feature values are generated, and the finger vein index feature value matching difference of a person to be recognized and a known person is calculated through a similarity measurement method; and the maximum matching difference value is screened, and whether the identities are matched or not is judged according to a set matching threshold value, so that high-precision identity authentication is realized. According to the invention, through multi-level feature fusion and accurate matching, the accuracy and reliability of finger vein identity verification are improved.

Description

Technical Field

[0001] The present invention relates to the field of biometric technologies, and particularly to a biometric recognition method based on finger vein image stratification. Background Art

[0002] As an important branch of biometric technologies, finger vein recognition has developed rapidly in recent years. Its core advantage lies in that the finger vein pattern is located under the human skin layer, which is difficult to replicate and not easily affected by external factors, such as stains or abrasions on the finger surface. Therefore, it has great advantages in terms of the security and stability of identity authentication. With the progress of computer vision, deep learning, and imaging technologies, finger vein recognition has gradually been applied to fields such as financial payment, access control, and medical record management.

[0003] Currently, finger vein recognition technology is developing towards high precision, low cost, and portability. Research mainly focuses on improving the acquisition quality of vein images, enhancing the robustness of feature extraction and matching algorithms, and further improving recognition accuracy through hierarchical analysis (such as the joint processing of shallow and deep veins). At the same time, with the popularization of mobile devices and Internet of Things technologies, biometric systems based on finger veins are also expanding towards the embedded and cross-platform directions, promoting the popular application of the technology.

[0004] In the prior art, traditional finger vein image recognition methods often have difficulty in effectively extracting subtle finger vein features, especially being interfered by image noise, resulting in low recognition accuracy. In the prior art, most methods rely on single image processing technologies, fail to fully utilize the multi-level features of finger vein images, and lack the comprehensive analysis of deep and surface vein features of the images. Therefore, how to efficiently denoise and extract the texture and geometric features of finger surface and deep veins, and then generate accurate finger vein feature values, is an urgent problem to be solved in the prior art;

[0005] Secondly, in the prior art, finger vein matching methods often only rely on simple similarity measurement methods, and do not fully consider the hierarchical structure of finger veins and their complex geometric features. This makes the matching process affected by noise and slight changes, and is prone to problems such as misrecognition and matching failure.

[0006] Therefore, it is necessary to provide a biometric recognition method based on finger vein image stratification to solve the above problems.

[0007] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0008] The object of the present invention is to provide a biometric recognition method based on finger vein image layering to solve the problems raised in the above-mentioned background technology.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] A biometric recognition method based on finger vein image layering, the specific steps include:

[0011] Step 1: Collect the finger vein image of the person to be recognized, preprocess the collected finger vein image to enhance the finger vein features and remove noise, and obtain the first recognition image of finger vein layering;

[0012] Step 2: Extract the region of interest of the finger in the first recognition image based on the edge detection algorithm, calculate the edge intensity of each edge pixel point in the region of interest, and set up a double-threshold detection to differentiate strong edge pixel points and weak edge pixel points, and combine adjacent strong edge pixels with the weak edge pixels connected to them to form a continuous edge curve, and divide the first recognition image into a feature region and a non-feature region;

[0013] Step 3: Layer the feature region of the finger vein image, extract the finger vein structure features of different levels, extract the texture features of the finger surface vein by the local binary algorithm to generate texture coefficient feature values, extract the geometric features of the finger deep vein by the high-pass filter to generate shape coefficient feature values, and use the generated texture coefficient feature values and shape coefficient feature values to generate the finger vein index feature value of the person to be recognized;

[0014] Step 4: Obtain the finger vein image information of known persons, generate the finger vein index feature values of known persons in the same way, calculate the matching difference value between the finger vein image of the person to be recognized and the finger vein index feature values of known persons by the similarity measurement method, and screen out the maximum matching difference value;

[0015] Step 5: Set a matching threshold. If the maximum matching difference value is less than or equal to the threshold, it is determined that the matching is successful and the identity confirmation is completed at the same time; if the maximum matching difference value is greater than the threshold, it is determined that the matching fails.

[0016] Further, the method for collecting the finger vein image of the person to be recognized and preprocessing the collected finger vein image to obtain the first recognition image of finger vein layering is as follows:

[0017] Adopt a near-infrared emission light source to make the deoxyhemoglobin in the internal vein of the finger absorb part of the infrared light to form a vein shadow, use a high-resolution camera to capture the vein shadow image, and ensure the stability of the finger position, light source angle and acquisition environment at the same time to reduce noise interference;

[0018] Grayscale the image, convert the color or infrared image into a grayscale image to reduce the computational complexity, and use Gaussian filtering to smooth the image to reduce noise interference, and finally obtain the first recognition image of finger vein stratification.

[0019] Furthermore, based on the edge detection algorithm, extract the region of interest of the finger, and calculate the edge intensity of each edge pixel in the region of interest. The method is as follows:

[0020] Apply the Sobel operator to each pixel in the first recognition image, where the center of the operator is aligned with the current pixel, and then multiply the corresponding elements to obtain the grayscale value of the new pixel. Determine the horizontal or vertical direction of each pixel by calculating the edge direction, and compare the grayscale values of adjacent pixels in the corresponding direction. If the pixel is not the local maximum in that direction, set its grayscale value to 0 to suppress non-edges. The logic is as follows:

[0021]

[0022]

[0023] Among them, A x represents the grayscale amplitude in the horizontal direction after the pixel is convolved by the operator, and A y represents the grayscale amplitude in the vertical direction after the pixel is convolved by the operator, a 11 ……a ij is the grayscale value of the pixel in the horizontal direction, b 11 ……b mn is the grayscale value of the pixel in the vertical direction, A (x，y) represents the edge intensity of the pixel, Q θ represents the edge direction of the pixel. i, j, m, and n are all positive integers, and i, j, m, and n are used to represent the indices of the pixel grayscale value matrix, that is, i and j are used to represent the coordinate indices of the pixels in the horizontal direction, and m and n are used to represent the coordinate indices of the pixels in the vertical direction.

[0024] Furthermore, divide the first recognition image into a feature region and a non-feature region. The method is as follows:

[0025] Double-threshold detection is set up, that is, two gradient thresholds are set to distinguish strong edge pixels from weak edge pixels. The high gradient component threshold is used to mark strong edge pixels, and pixels exceeding this threshold are marked as strong edge pixels; the low gradient component threshold is used to mark weak edge pixels, and pixels below this threshold are marked as weak edge pixels; pixels between the two are marked as non-edge pixels. Starting from the strong edge pixels, the edge tracking technique is used to combine adjacent strong edge pixels with the weak edge pixels connected to them to form a continuous edge curve. The edge curve is used as a dividing line to distinguish the feature area from the non-feature area, and the detected area is classified and recognized according to the following logic:

[0026]

[0027] Among them, T max and T min represent the high gradient component threshold and the low gradient component threshold respectively, and A q , A f , A r represent strong edge pixels, non-edge pixels, and weak edge pixels respectively.

[0028] Furthermore, the feature area of the finger vein image is stratified, and the finger vein structure features at different levels are extracted according to the following method:

[0029] The texture features of the finger surface veins are extracted through the local binarization algorithm to generate texture coefficient eigenvalues. The area formed by the cycloid with the central pixel as the center and the neighborhood radius is defined as the neighborhood. Eight neighborhood pixel numbers are selected to compare with the gray value of the central pixel. If the neighborhood pixel value is greater than or equal to the gray value of the central pixel, the corresponding binary bit is set to 1, otherwise it is set to 0. The obtained binary string is combined clockwise into a binary number, and after decimal conversion, it represents the texture coefficient eigenvalue of the defect area according to the following logic:

[0030]

[0031] Among them, C l represents the gray value of the l-th neighborhood pixel, l takes values from 1 to 8 in turn, C z represents the gray value of the central pixel, and M(l) represents the judgment value of the gray scale of the l-th neighborhood pixel. When M(l)=1, it means that the gray value of the l-th neighborhood pixel is greater than the gray value of the central pixel; when M(l)=0, it means that the gray value of the l-th neighborhood pixel is less than the gray value of the central pixel;

[0032] Using the obtained binary number (M 1 M 2 M 3 M4 M 5 M 6 M 7 M 8 )Perform decimal conversion to obtain a decimal conversion number M', and this decimal number M' is the texture coefficient eigenvalue of the finger vein.

[0033] Extract the geometric features of the deep finger vein through a high-pass filter to generate the shape coefficient eigenvalue. Extracting the geometric features of the deep finger vein includes the number of finger vein branch points, the number of finger vein main intersection points, the length and width of the finger vein. Calculate the ratio of the number of finger vein main intersection points to the number of finger vein branch points, and the ratio of the length and width of the finger vein to obtain the finger vein bifurcation ratio and the finger vein length-width ratio. Based on the finger vein bifurcation ratio and the finger vein length-width ratio, generate the shape coefficient eigenvalue representing the shape complexity and topological structure characteristics of the finger vein. The formula is as follows:

[0034]

[0035] Among them, VBR and VTR are the finger vein bifurcation ratio and the finger vein length-width ratio respectively, V is the shape coefficient eigenvalue of the finger vein, N 1 , N 2 are the number of finger vein branch points and the number of finger vein main intersection points, k 1 , k 2 are the length and width of the finger vein.

[0036] Furthermore, use the generated texture coefficient eigenvalue and shape coefficient eigenvalue to comprehensively generate a unique finger vein index eigenvalue. The formula is as follows:

[0037]

[0038] Among them, Tape represents the unique finger vein index eigenvalue of the finger vein image of the person to be identified.

[0039] Furthermore, use the similarity measurement method to calculate the matching difference value between the finger vein image of the person to be identified and the finger vein index eigenvalue among known persons, and screen out the maximum matching difference value. The method is as follows:

[0040] Based on the unique finger vein index eigenvalue generated in step 3, use the Euclidean distance algorithm to measure the difference degree between the unique finger vein index eigenvalue of the image of the person to be identified and the finger vein index eigenvalue of the known person. The formula is as follows:

[0041] D E =MAX|(Tape - F r )|

[0042] Among them, D EF is the maximum degree of difference in features between the finger vein index feature value of the person image to be recognized and the finger vein index feature value of the known person image, that is, the maximum matching difference value. r is the finger vein index feature value of the r-th known person, Tape is the only finger vein index feature value in the finger vein image of the person to be recognized, r is the index of the known person, r = 1, 2, 3…n, and n is the specific number of known persons.

[0043] Furthermore, a matching threshold is set to determine whether the biometric recognition is successfully matched. The logical formula is as follows:

[0044]

[0045] Among them, Q represents the logical value for judging whether the biometric recognition is successfully matched. When Q = 0, the maximum matching difference value is greater than the set threshold, indicating low similarity and failed matching; when Q = 1, the maximum matching difference value is less than or equal to the set threshold, indicating high similarity and successful matching.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0047] Through the hierarchical processing and multi-level feature extraction of finger vein images, the present invention can effectively improve the extraction accuracy of finger vein features. In particular, by combining the texture features of superficial veins and the geometric features of deep veins, and through methods such as local binarization and high-pass filters, the problem of image noise interference is solved, making the vein features in the image clearer and more detailed, thereby improving the accuracy of finger vein recognition. This hierarchical processing method can more comprehensively capture various detailed information in the vein image, making the feature values more unique and providing a more reliable basis for subsequent matching;

[0048] Secondly, by introducing double-threshold detection, edge tracking and similarity measurement methods, the present invention effectively improves the stability and accuracy of the matching process. Using the Euclidean distance algorithm to calculate the difference between finger vein index feature values and combining the set matching threshold can accurately judge the matching situation of the recognition result and avoid misrecognition and false positive problems. Through these technical means, high recognition accuracy can be maintained under different environments and conditions, ensuring the efficiency and robustness of biometric recognition, especially suitable for identity verification tasks in complex environments in practical applications;

[0049] Through the hierarchical processing, feature extraction and similarity measurement methods of finger vein images, the present invention realizes high-precision and high-stability biometric recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0051] To make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to specific embodiments.

[0052] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. The terms such as "comprising" or "including" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left" and "right" are only used to represent relative position relationships, and when the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0053] Embodiment:

[0054] Please refer to Figure 1 , the present invention provides a technical solution:

[0055] A biometric recognition method based on finger vein image layering, and the specific steps include:

[0056] Step 1: Collect the finger vein image of the person to be recognized, preprocess the collected finger vein image to enhance the finger vein features and remove noise, and obtain the first recognition image of finger vein layering;

[0057] Step 2: Extract the region of interest of the finger in the first recognition image based on the edge detection algorithm, calculate the edge intensity of each edge pixel point in the region of interest, and set up a double-threshold detection to distinguish strong edge pixel points and weak edge pixel points, combine adjacent strong edge pixels with the weak edge pixels connected to them to form a continuous edge curve, and segment the first recognition image into a feature region and a non-feature region;

[0058] Step 3: Layer the feature region of the finger vein image, extract the finger vein structure features of different levels, extract the texture features of the finger surface vein by the local binary algorithm to generate texture coefficient feature values, extract the geometric features of the finger deep vein by the high-pass filter to generate shape coefficient feature values, and use the generated texture coefficient feature values and shape coefficient feature values to generate the finger vein index feature value of the person to be recognized;

[0059] Step 4: Obtain the finger vein image information of known individuals, generate the finger vein index feature values of known individuals according to the same method, calculate the matching difference value between the finger vein image of the person to be identified and the finger vein index feature values of known individuals using a similarity measurement method, and screen out the maximum matching difference value;

[0060] Step 5: Set a matching threshold. If the maximum matching difference value is less than or equal to the threshold, it is determined that the matching is successful, and the identity confirmation is completed at the same time; if the maximum matching difference value is greater than the threshold, it is determined that the matching fails.

[0061] It should be noted that through the cooperation of a near-infrared light source and a high-resolution camera, the venous shadow image is effectively captured, ensuring the clear presentation of venous features in the image. By stabilizing the finger position, light source angle, and acquisition environment, noise interference is reduced, and the image quality is improved. Grayscale conversion and Gaussian filtering further simplify the image data, reduce the computational complexity, and remove unnecessary noise at the same time, providing a high-quality first recognition image for subsequent hierarchical feature extraction and recognition, which is the basis for ensuring the high efficiency and accuracy of the finger vein recognition system.

[0062] Among them, the area where the finger is collected is the region of interest. Generally, it is selected in the area where the finger vein distribution is relatively rich, such as the finger pulp, finger side, and fingertip. The finger vein distribution in the finger pulp, finger side, and fingertip regions of the finger is relatively rich. The veins in these parts are usually relatively superficial, and the blood flow rate is moderate. This makes it easier to capture clear and accurate vein textures when collecting finger vein images or vascular information in these areas, reducing interference during collection. Moreover, there are usually no excessive skin folds or other interference factors in these areas of the finger, making the vein morphology more obvious. The finger vein recognition technology can utilize these features, and the images obtained from these parts are easier to distinguish from the vascular images of other parts, reducing the possibility of misidentification.

[0063] It should be noted that known individuals can be any individuals whose biometric data has been registered in the finger vein recognition system, covering various groups such as enterprise employees, government personnel, financial customers, and ordinary consumers. These data are usually bound to identity information to ensure the accuracy and security of the recognition results.

[0064] Therefore, it is necessary to collect the finger vein image of the person to be identified, preprocess the collected finger vein image, and obtain the first recognition image of finger vein stratification. The method is as follows:

[0065] Adopt a near-infrared emission light source to make the deoxyhemoglobin in the internal veins of the finger absorb part of the infrared light to form a venous shadow, use a high-resolution camera to capture the venous shadow image, and ensure the stability of the finger position, light source angle, and acquisition environment at the same time to reduce noise interference;

[0066] Gray-scale the image, convert the color or infrared image into a grayscale image to reduce the computational complexity, and use Gaussian filtering to smooth the image to reduce noise interference, and finally obtain the first recognition image of finger vein stratification.

[0067] It should be noted that using the Sobel operator for edge detection is an important step in finger vein image processing because it can accurately extract the edge information in the image, especially in the region of interest of the finger. By calculating the gray-scale changes of each pixel point in the horizontal and vertical directions, the Sobel operator effectively enhances the saliency of the vein edges, so that the vein contours can be distinguished more clearly. Through local maximum suppression, the interference of non-edge regions is further removed, ensuring the accuracy of edge detection. The calculation of this edge strength and direction lays the foundation for subsequent feature extraction and classification recognition, ensuring the efficiency and stability of the system in practical applications.

[0068] Therefore, it is necessary to extract the region of interest of the finger based on the edge detection algorithm and calculate the edge strength of each edge pixel point in the region of interest. The method is as follows:

[0069] Apply the Sobel operator to each pixel point in the first recognition image, where the center of the operator is aligned with the current pixel, and then multiply the corresponding elements to obtain the gray-scale value of the new pixel point. Determine the horizontal or vertical direction of each pixel point by calculating the edge direction, and compare the gray-scale values of adjacent pixels in the corresponding direction. If the pixel point is not the local maximum in that direction, set its gray-scale value to 0 to suppress non-edges. The logic is as follows:

[0070]

[0071] Among them, A x represents the gray-scale amplitude of the pixel point in the horizontal direction after convolution by the operator, A y represents the gray-scale amplitude of the pixel point in the vertical direction after convolution by the operator, a 11 ……a ij is the gray-scale value of the pixel point in the horizontal direction, b 11 ……b mn is the gray-scale value of the pixel point in the vertical direction, A (x，y) represents the edge strength of the pixel point, Q θ represents the edge direction of the pixel point. i, j, m, and n are all positive integers, and i, j, m, and n are used to represent the indices of the pixel point gray-scale value matrix, that is, i and j are used to represent the coordinate indices of the pixel point in the horizontal direction, and m and n are used to represent the coordinate indices of the pixel point in the vertical direction.

[0072] It should be noted that the importance of the double-threshold detection method in image segmentation lies in its ability to effectively distinguish strong edges, weak edges, and non-edge pixel points in an image, ensuring the accurate extraction of the venous feature region. By setting a high gradient threshold and a low gradient threshold, the key edge regions in the image can be clearly identified, and the edge tracking technology further ensures the coherence of adjacent strong-edge pixels and weak-edge pixels, forming a continuous edge curve. This method effectively reduces noise interference, ensures the precise division of the feature region, provides a reliable basis for subsequent detection and classification recognition, and enhances the robustness and accuracy of the recognition process.

[0073] Therefore, the first recognition image needs to be segmented into a feature region and a non-feature region, and the method is as follows:

[0074] Set up double-threshold detection, that is, set two gradient thresholds to distinguish strong-edge pixel points and weak-edge pixel points. The high-gradient component threshold is used to mark strong-edge pixel points, and pixel points exceeding this threshold are marked as strong-edge pixel points; the low-gradient component threshold is used to mark weak-edge pixel points, and pixel points below this threshold are marked as weak-edge pixel points; pixel points between the two are marked as non-edge pixel points. Starting from the strong-edge pixel points, use edge tracking technology to combine adjacent strong-edge pixels with the weak-edge pixels connected to them to form a continuous edge curve. Use the edge curve as a boundary to distinguish the feature region and the non-feature region, and classify and recognize the detected regions. The logic is as follows:

[0075]

[0076] Among them, T max 、T min represent the high-gradient component threshold and the low-gradient component threshold respectively, and A q 、A f 、A r represent strong-edge pixel points, non-edge pixel points, and weak-edge pixel points respectively.

[0077] It should be noted that the hierarchical extraction of the structural features of finger veins is crucial for improving the recognition accuracy. Since the superficial and deep veins of the finger each contain different information and features, they can provide more abundant finger vein authentication information. Extracting the texture features of the superficial finger veins through the local binary pattern algorithm helps to capture the subtle texture differences of the veins, and the generated texture coefficient feature values provide a quantitative basis for further feature classification. At the same time, by using a high-pass filter to extract the geometric features of the deep veins and calculating the bifurcation ratio and aspect ratio of the finger veins, the shape complexity and topological structure characteristics of the veins can be characterized, providing higher-dimensional geometric information for vein recognition. The extraction of these hierarchical features not only enhances the distinguishability of finger vein images but also provides a more accurate and comprehensive basis for subsequent classification and recognition, thus improving the accuracy and robustness of the system.

[0078] Therefore, it is necessary to layer the feature regions of finger vein images and extract the structural features of finger veins at different levels. The method is as follows:

[0079] Extract the texture features of the superficial finger veins through the local binary pattern algorithm to generate texture coefficient feature values. Taking the central pixel as the center and the neighborhood radius as the cycloid to form a region as the neighborhood, select eight neighborhood pixel numbers to compare with the gray value of the central pixel point. If the neighborhood pixel value is greater than or equal to the gray value of the central pixel point, the corresponding binary bit is set to 1; otherwise, it is set to 0. Combine the obtained binary string clockwise into a binary number, and after decimal conversion, it represents the texture coefficient feature value of this region. The logic is as follows:

[0080]

[0081] Among them, C l represents the gray value of the l-th neighborhood pixel point, where l takes values from 1 to 8 in sequence, C z represents the gray value of the central pixel point, and M(l) represents the judgment value of the gray scale of the l-th neighborhood pixel point. When M(l) = 1, it means that the gray value of the l-th neighborhood pixel point is greater than the gray value of the central pixel point; when M(l) = 0, it means that the gray value of the l-th neighborhood pixel point is less than the gray value of the central pixel point.

[0082] Use the obtained binary number (M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 ) for decimal conversion to obtain a decimal conversion number M'. This decimal number M' is the texture coefficient feature value of the finger vein;

[0083] Extract the geometric features of the deep finger veins through a high-pass filter to generate shape coefficient eigenvalues. The geometric features of the deep finger veins include the number of finger vein branch points, the number of main finger vein intersection points, the length and width of the finger veins. Calculate the ratio of the number of main finger vein intersection points to the number of finger vein branch points, and the ratio of the length and width of the finger veins to obtain the finger vein bifurcation ratio and the finger vein length-width ratio. According to the finger vein bifurcation ratio and the finger vein length-width ratio, generate shape coefficient eigenvalues representing the shape complexity and topological structure characteristics of the finger veins. The formula is as follows:

[0084]

[0085] Among them, VBR and VTR are the finger vein bifurcation ratio and the finger vein length-width ratio respectively, V is the shape coefficient eigenvalue of the finger vein, N 1 , N 2 are the number of finger vein branch points and the number of main finger vein intersection points, k 1 , k 2 are the length and width of the finger vein.

[0086] It should be noted that the texture features of the superficial veins and the geometric shape features of the deep veins are organically combined to form a comprehensive and unique identifier. By fusing the texture coefficient eigenvalues and the shape coefficient eigenvalues, the finger vein index eigenvalue can make full use of the multi-level and multi-dimensional information of the veins, thus significantly improving the discrimination ability and accuracy of the recognition system. This eigenvalue not only combines texture and geometric information, but also simplifies the expression form of the features through mathematical formulas, providing a compact, effective and operable feature representation, ensuring the efficiency and reliability of the finger vein in the authentication process.

[0087] Therefore, it is necessary to comprehensively generate a unique finger vein index eigenvalue by using the generated texture coefficient eigenvalue and shape coefficient eigenvalue. The formula is as follows:

[0088]

[0089] Among them, Tape represents the finger vein index eigenvalue.

[0090] It should be noted that by calculating the feature difference between the image to be recognized and the database template, the Euclidean distance can accurately reflect the similarity between the two. The smaller the matching value, the more similar the image to be recognized is to the database template, and the higher the recognition accuracy. This method effectively converts the texture and geometric features of the finger vein into numerical features, and through a simple and effective distance metric, realizes a fast and efficient matching process. It ensures that the finger vein authentication system can efficiently retrieve in a large-scale database and has high recognition accuracy and stability.

[0091] Therefore, it is necessary to use a similarity measurement method to calculate the matching difference value between the finger vein image of the person to be identified and the finger vein index eigenvalue of the known person, and screen out the maximum matching difference value. The method is as follows:

[0092] Based on the unique finger vein index eigenvalue generated in step 3, the Euclidean distance algorithm is used to measure the difference degree between the unique finger vein index eigenvalue of the image of the person to be identified and the finger vein index eigenvalue of the known person. The formula is as follows:

[0093] D E =MAX|(Tape - F r )|

[0094] where D E is the maximum characteristic difference degree between the finger vein index eigenvalue of the image of the person to be identified and the finger vein index eigenvalue of the image of the known person, that is, the maximum matching difference value. F r is the finger vein index eigenvalue of the rth known person. Tape is the unique finger vein index eigenvalue in the finger vein image of the person to be identified. r is the index of the known person, r = 1, 2, 3...n, and n is the specific number of known persons.

[0095] It should be noted that by setting an appropriate matching threshold, the system can effectively distinguish between successful and failed matches. When the matching value of the image to be identified and the database template is less than or equal to the threshold, the system considers that the image similarity is high enough and the verification passes; otherwise, when the matching value is greater than the threshold, the similarity is low and the recognition fails. This method ensures the accuracy and robustness of the finger vein recognition system, can avoid misrecognition or missed recognition, and guarantees the accuracy and security of the identity verification process. The setting of the threshold has an important impact on the sensitivity and error rate of the system. A reasonable threshold design can optimize the recognition effect and improve the reliability and efficiency of the system in practical applications.

[0096] Therefore, it is necessary to set a matching threshold to determine whether the biometric recognition is successfully matched. The logical formula is as follows:

[0097]

[0098] where Q represents the logical value for judging whether the biometric recognition is successfully matched. When Q = 0, the maximum matching difference value is greater than the set threshold, indicating low similarity and failed matching; when Q = 1, the maximum matching difference value is less than or equal to the set threshold, indicating high similarity and successful matching.

[0099] The above formulas are all dimensionless and take their numerical calculations. The formula is obtained by software simulation of a large amount of collected data to get a formula closest to the real situation. The preset parameters in the formula are set by technicians in the field according to the actual situation.

[0100] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0101] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. They may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0102] As mentioned above, the above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application.

Claims

1. A biometric feature recognition method based on finger vein image layering, characterized in that: The specific steps include: Step 1: Collect the finger vein image of the person to be identified, and pre-process the collected finger vein image to enhance the finger vein features and remove noise, so as to obtain the first identification image of finger vein layering; Step 2: Based on the edge detection algorithm, the region of interest of the finger in the first recognition image is extracted, the edge strength of each edge pixel in the region of interest is calculated, and a double threshold detection is established to differentiate strong edge pixels and weak edge pixels, and the adjacent strong edge pixels are combined with the weak edge pixels connected to them to form a continuous edge curve, and the first recognition image is segmented into a feature region and a non-feature region; Step 3: Layer the characteristic areas of the finger vein image, extract the finger vein structure features at different levels, extract the texture features of the surface veins of the finger through the local binarization algorithm to generate the texture coefficient eigenvalue, extract the geometric features of the deep veins of the finger through the high-pass filter to generate the shape coefficient eigenvalue, and use the generated texture coefficient eigenvalue and shape coefficient eigenvalue to generate the finger vein index eigenvalue of the person to be identified; Step 4: Obtain the finger vein image information of the known person, generate the finger vein index feature value of the known person in the same way, calculate the matching difference value between the finger vein image of the person to be identified and the finger vein index feature value of the known person by using the similarity measurement method, and select the maximum matching difference value; Step 5: Set the matching threshold. If the maximum matching difference value is less than or equal to the threshold, the match is considered successful and identity confirmation is completed. If the maximum matching difference value is greater than the threshold, the match is considered unsuccessful.

2. The biometric feature recognition method based on finger vein image layering according to claim 1 is characterized in that: The finger vein image of the person to be identified is collected, and the collected finger vein image is preprocessed to obtain a first identification image of finger vein layering, according to the method: Adopt near-infrared emission light source to make the deoxyhemoglobin in the veins inside the finger absorb part of the infrared light to form a vein shadow. Use a high-resolution camera to capture the vein shadow image while ensuring the stability of the finger position, light source angle and acquisition environment to reduce noise interference. Grayscale the image, convert the color or infrared image into a grayscale image to reduce the computational complexity, use Gaussian filtering to smooth the image, reduce noise interference, and finally obtain the first recognition image of the finger vein stratification.

3. The biometric feature recognition method based on finger vein image layering according to claim 1 is characterized in that: The region of interest of the finger is extracted based on the edge detection algorithm, and the edge strength of each edge pixel in the region of interest is calculated. The method is based on: The Sobel operator is applied to each pixel in the first recognition image, where the center of the operator is aligned with the current pixel, and then the corresponding elements are multiplied to obtain the new pixel grayscale value. The horizontal or vertical direction of each pixel is determined by calculating the edge direction, and the grayscale values ​​of adjacent pixels in the corresponding direction are compared. If the pixel is not the local maximum in this direction, its grayscale value is set to 0 to suppress non-edges. The logic is as follows: Among them, A x A represents the grayscale amplitude of the pixel in the horizontal direction after the operator convolution. y Represents the grayscale amplitude of the pixel in the vertical direction after operator convolution, a 11 ……a ij is the gray value of the pixel in the horizontal direction, b 11 ……b mn is the gray value of the pixel in the vertical direction, A (x，y) Indicates the edge strength of the pixel, Q θ Indicates the edge direction of the pixel, i, j, m, and n are all positive integers, and i, j, m, and n are the indices used to represent the grayscale value matrix of the pixel, that is, i and j are the coordinate indices used to represent the pixel in the horizontal direction, and m and n are the coordinate indices used to represent the pixel in the vertical direction.

4. The biometric feature recognition method based on finger vein image layering according to claim 1 is characterized in that ,The first recognition image is divided into feature areas and non-feature areas according to the following method: A dual threshold detection is established, that is, two gradient thresholds are established to distinguish strong edge pixels from weak edge pixels. The high gradient component threshold is used to mark strong edge pixels, and pixels exceeding this threshold are marked as strong edge pixels; the low gradient component threshold is used to mark weak edge pixels, and pixels below this threshold are marked as weak edge pixels; pixels between the two are marked as non-edge pixels. Starting from the strong edge pixels, the edge tracking technology is used to combine the adjacent strong edge pixels with the weak edge pixels connected to them to form a continuous edge curve. The edge curve is used as a dividing line to distinguish the feature area and the non-feature area, and the detected area is classified and identified. The logic is as follows: Among them, T max , T min Respectively represent the high gradient component threshold and the low gradient component threshold, A q , A f , A r They represent strong edge pixels, non-edge pixels, and weak edge pixels respectively.

5. The biometric feature recognition method based on finger vein image layering according to claim 1 is characterized in that ,The characteristic areas of the finger vein image are layered, and the finger vein structure features at different levels are extracted. The method is based on: The texture features of the superficial veins of the finger are extracted by the local binarization algorithm to generate the texture coefficient eigenvalue. The neighborhood is defined as the area formed by the cycloid with the central pixel as the center and the neighborhood radius as the neighborhood. Eight neighborhood pixels are selected to compare with the gray value of the central pixel. If the neighborhood pixel value is greater than or equal to the gray value of the central pixel, the binary bit at the corresponding position is set to 1, otherwise it is set to 0. The obtained binary string is combined into a binary number in a clockwise direction. After decimal conversion, the texture coefficient eigenvalue of the defect area is characterized. The logic is as follows: Among them, C l represents the gray value of the lth neighborhood pixel, l ranges from 1 to 8, C z represents the gray value of the central pixel, M(l) represents the judgment value of the gray value of the lth neighboring pixel. When M(l) = 1, it means that the gray value of the lth neighboring pixel is greater than the gray value of the central pixel; when M(l) = 0, it means that the gray value of the lth neighboring pixel is less than the gray value of the central pixel. The obtained binary number (M1M2M3M4M5M6M7M8) is used to perform decimal conversion to obtain a decimal conversion number M', and the decimal number M' is the texture coefficient eigenvalue of the finger vein; The geometric features of the deep veins of the fingers are extracted by a high-pass filter to generate shape coefficient eigenvalues. The geometric features of the deep veins of the fingers are extracted, including the number of branch points of the finger veins, the number of main intersection points of the finger veins, the length and width of the finger veins. The ratio of the number of main intersection points of the finger veins to the number of branch points of the finger veins, and the ratio of the length and width of the finger veins are calculated to obtain the finger vein bifurcation ratio and the finger vein length-width ratio. According to the finger vein bifurcation ratio and the finger vein length-width ratio, the shape coefficient eigenvalues ​​characterizing the shape complexity and topological structure characteristics of the finger veins are generated. The formula is as follows: Among them, VBR and VTR refer to the bifurcation ratio and length-to-width ratio of the finger vein, respectively; V refers to the characteristic value of the shape coefficient of the vein; N1 and N2 refer to the number of venous branch points and the number of main intersections of the finger vein; k1 and k2 refer to the length and width of the vein.

6. The biometric feature recognition method based on finger vein image layering according to claim 5 is characterized in that: The generated texture coefficient eigenvalue and shape coefficient eigenvalue are used to comprehensively generate a unique finger vein index eigenvalue, based on the following formula: Wherein, Tape represents the unique finger vein index feature value of the finger vein image of the person to be identified.

7. The biometric feature recognition method based on finger vein image layering according to claim 1 is characterized in that: The similarity measurement method is used to calculate the matching difference between the finger vein image of the person to be identified and the finger vein index feature value of the known person, and the maximum matching difference value is screened out. The method is based on: Based on the unique finger vein index feature value generated in step 3, the Euclidean distance algorithm is used to measure the difference between the unique finger vein index feature value of the image of the person to be identified and the finger vein index feature value of the known person, based on the formula: D E =MAX|(Tape-F r )| Among them, D E is the maximum difference between the finger vein index feature value of the image of the person to be identified and the finger vein index feature value of the known person image, that is, the maximum matching difference value, F r is the finger vein index feature value of the rth known person, Tape is the unique finger vein index feature value in the finger vein image of the person to be identified, r is the index of the known person, r = 1, 2, 3…n, n is the specific number of known persons.

8. The biometric feature recognition method based on finger vein image layering according to claim 1 is characterized in that: The matching threshold is set to determine whether the biometric feature recognition match is successful. The logic formula is as follows: Among them, Q represents the logical value for judging whether the biometric feature is successfully identified and matched. When Q=0, the maximum matching difference value is greater than the set threshold, indicating that the similarity is low and the match fails; when Q=1, the maximum matching difference value is less than or equal to the set threshold, indicating that the similarity is high and the match is successful.

Images