Finger vein recognition method, device, computer equipment and storage medium
By constructing a filter based on the growth characteristics of finger veins, region cropping and feature extraction of finger vein images are performed, solving the accuracy problem of traditional finger vein recognition methods and achieving more efficient recognition results.
Patent Information
- Application Number
- CN202011294170.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-11-18
- Publication Date
- 2024-11-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In traditional finger vein recognition methods, general filters can only extract specific aspects of finger vein features from finger vein images, leading to inaccurate recognition.
A filter constructed using information on finger vein growth characteristics is used to perform region cropping and feature extraction of finger vein images, generating finger vein features corresponding to the finger vein growth characteristics information, and comparing them with preset features to determine the recognition result.
It improves the accuracy of finger vein recognition and the efficiency of feature extraction, and enhances the adaptability to finger vein patterns.
Smart Images

Figure CN112395988B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to a finger vein recognition method, apparatus, computer equipment and storage medium. Background Art
[0002] Finger vein recognition refers to the use of the distribution image of the vein network in the finger, that is, the finger vein image, for identity recognition. As one of the representative technologies in identity recognition, finger vein recognition has the advantages of non-contact collection, liveness recognition, internal features, small size, and simple imaging device structure. It has received widespread attention in the field of biometric recognition. In the traditional method, a universal filter is used to extract the finger vein features in the finger vein image to achieve finger vein recognition.
[0003] However, the general filter used in the traditional method can only extract specific aspects of finger vein features in the finger vein image. Since finger vein recognition will be affected by many aspects of the finger vein image, the traditional method cannot accurately perform finger vein recognition. Summary of the invention
[0004] Based on this, it is necessary to provide a finger vein recognition method, device, computer equipment and storage medium that can improve the accuracy of finger vein recognition in response to the above technical problems.
[0005] A finger vein recognition method, the method comprising:
[0006] Obtaining a finger vein image to be identified;
[0007] Performing regional interception on the finger vein image to be identified to obtain a target region image;
[0008] Calling a pre-built filter, where the filter is built based on finger vein growth characteristic information;
[0009] Extracting features of the target area image by using the filter to obtain finger vein features corresponding to the finger vein growth characteristic information;
[0010] Comparing the finger vein feature with the preset feature to obtain feature similarity;
[0011] A preset feature whose feature similarity is greater than a threshold is selected, and the identity information corresponding to the preset feature is determined as a finger vein recognition result.
[0012] In one embodiment, extracting features from the target area image through the filter to obtain finger vein features corresponding to the finger vein growth characteristic information includes:
[0013] Dividing the target area image into a plurality of units by using the filter;
[0014] Calculate a first amplitude response map and a first frequency direction response map corresponding to the finger vein growth characteristic information according to the pixel points in each unit;
[0015] A target histogram is generated according to the first amplitude response graph and the first frequency direction response graph, and the target histogram is determined to be a finger vein feature corresponding to the finger vein growth characteristic information.
[0016] In one embodiment, the calculating of the first amplitude response map and the first frequency direction response map corresponding to the finger vein growth characteristic information according to the pixel points in each unit includes:
[0017] Calculate the filter response map corresponding to each pixel point in each unit and in accordance with the finger vein growth characteristic information;
[0018] A first amplitude response graph and a first frequency direction response graph corresponding to the finger vein growth characteristic information are calculated according to the filter response graph.
[0019] In one embodiment, generating a target histogram according to the first amplitude response graph and the first frequency direction response graph, and determining that the target histogram is a finger vein feature corresponding to the finger vein growth characteristic information includes:
[0020] generating a unit histogram according to the first amplitude response graph and the first frequency direction response graph;
[0021] Combining adjacent unit histograms according to preset parameters to generate a block histogram;
[0022] Normalizing the block histogram to obtain a standard histogram;
[0023] The standard histograms are combined to obtain the finger vein features corresponding to the finger vein growth characteristic information.
[0024] In one embodiment, before acquiring the finger vein image to be identified, the method further includes:
[0025] Acquire multiple finger vein sample images;
[0026] Performing region interception on multiple finger vein images to obtain multiple region images to be processed;
[0027] The finger vein growth characteristic information is obtained by calculating the pixel points in the plurality of images of the area to be processed;
[0028] A filter is constructed according to the finger vein growth characteristic information.
[0029] In one embodiment, the step of calculating the finger vein growth characteristic information based on the pixel points in the plurality of images of the area to be processed includes:
[0030] Calculating a second amplitude response map and a second frequency direction response map corresponding to the corresponding image of the region to be processed according to the pixel points in each image of the region to be processed, to obtain a plurality of second amplitude response maps and a plurality of second frequency response maps;
[0031] The finger vein growth characteristic information is calculated based on the multiple second amplitude response graphs and the multiple second frequency direction response graphs.
[0032] In one embodiment, the calculating of the finger vein growth characteristic information according to the plurality of second amplitude response graphs and the plurality of second frequency direction response graphs includes:
[0033] Divide each image of the area to be processed into a plurality of units;
[0034] Accumulating a plurality of second amplitude response graphs corresponding to the same unit according to a preset frequency and a preset direction to obtain a cumulative histogram corresponding to each unit;
[0035] Select the target frequency and target direction in the cumulative histogram corresponding to each unit to obtain the target frequency and target direction corresponding to each unit;
[0036] Finger vein growth characteristic information is generated according to target frequencies and target directions corresponding to the multiple units.
[0037] A finger vein recognition device, comprising:
[0038] Image acquisition module, voice acquisition of finger vein images to be identified;
[0039] A region interception module, used for performing region interception on the finger vein image to be identified to obtain a target region image;
[0040] A calling module, used for calling a pre-built filter, wherein the filter is built based on the finger vein growth characteristic information;
[0041] A feature extraction module, used to extract features of the target area image through the filter to obtain finger vein features corresponding to the finger vein growth characteristic information;
[0042] A feature comparison module, used to compare the finger vein feature with a preset feature to obtain feature similarity;
[0043] The result determination module is used to select the preset features whose feature similarity is greater than a threshold value, and determine the identity information corresponding to the preset features as the finger vein recognition result.
[0044] A computer device includes a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and the processor implements the steps in the above-mentioned various method embodiments when executing the computer program.
[0045] A computer-readable storage medium stores a computer program, which implements the steps in the above-mentioned method embodiments when executed by a processor.
[0046] The finger vein recognition method, device, computer equipment and storage medium obtain the finger vein image to be identified, perform regional interception on the finger vein image to be identified, obtain the target area image, and then call the pre-constructed filter. The filter is constructed based on the finger vein growth characteristic information. The filter extracts the features of the target area image to obtain the finger vein features corresponding to the finger vein growth characteristic information. Then, the finger vein features are compared with the preset features to obtain the feature similarity, and the preset features with feature similarity greater than the threshold are selected to determine the identity information corresponding to the preset features as the finger vein recognition result. Since the filter is constructed based on the finger vein growth characteristic information, the finger vein features can be extracted according to the distribution law of the frequency and direction of the finger veins in different finger regions obtained by statistics, so that the finger vein features are more consistent with the finger vein generation characteristic information. Compared with the traditional general filter that can only extract the finger vein features of specific aspects in the finger vein image, more effective and accurate finger vein features can be extracted, thereby improving the accuracy of finger vein recognition. At the same time, the adaptability of the filter to the finger vein patterns in the finger vein features and the extraction efficiency of the finger vein features are also improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 A diagram showing an application environment of a finger vein recognition method in an embodiment;
[0048] Figure 2 A schematic diagram of a flow chart of a finger vein recognition method in one embodiment;
[0049] Figure 3 A schematic flow chart of a step of extracting features of a target area image through a filter to obtain finger vein features corresponding to finger vein growth characteristic information in one embodiment;
[0050] Figure 4 A schematic flow chart of a step of generating a target histogram according to a first amplitude response graph and a first frequency direction response graph and determining that the target histogram is a finger vein feature corresponding to finger vein growth characteristic information in an embodiment;
[0051] Figure 5 A schematic diagram of a flow chart of a filter construction step in one embodiment;
[0052] Figure 6 is a schematic diagram of the distribution of finger veins in a target direction in one embodiment;
[0053] Figure 7 is a structural block diagram of a finger vein recognition device in one embodiment;
[0054] Figure 8 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION
[0055] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0056] The finger vein recognition method provided in this application can be applied to Figure 1 In the application environment shown. Among them, the image acquisition device 102 communicates with the server 104 through the network. When finger vein recognition is required, the image acquisition device can send the acquired finger vein image to be recognized to the server 102. After acquiring the finger vein image to be recognized, the server 104 performs regional interception on the finger vein image to be recognized to obtain the target area image, thereby calling a pre-constructed filter, which is constructed based on the finger vein growth characteristic information. The filter is used to extract features of the target area image to obtain the finger vein features corresponding to the finger vein growth characteristic information, and then the finger vein features are compared with the preset features to obtain feature similarity, and the preset features with feature similarity greater than the threshold are selected, and the identity information corresponding to the preset features is determined as the finger vein recognition result. Among them, the image acquisition device can be a camera, a video camera, etc. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.
[0057] In one embodiment, Figure 2 As shown, a finger vein recognition method is provided, which is applied to Figure 1 The server in the example is used to illustrate the following steps:
[0058] Step 202, obtaining a finger vein image to be identified.
[0059] The finger vein image to be identified refers to an image that needs to be identified by finger vein recognition to achieve identity recognition. The finger vein image to be identified is used to represent the vein network distribution in the finger.
[0060] In scenarios where identity recognition is required, such as access control and attendance, the finger vein recognition image to be identified can be collected by an image acquisition device. When the user places the finger in the image acquisition area of the image acquisition device, the image acquisition device collects the corresponding finger vein image to be identified. The finger vein image to be identified collected by the image acquisition device can be a grayscale image without color change. The server can thus obtain the vein image to be identified sent by the image acquisition device.
[0061] Step 204 , performing region capture on the finger vein image to be identified to obtain a target region image.
[0062] The target area image refers to an image corresponding to the region of interest (ROI) in the finger vein image to be identified. The region of interest can be obtained by extracting the main part of the finger vein image, removing the background and edge parts, and performing normalization processing. The main part can be the middle part between the fingertip and the base of the finger.
[0063] Specifically, the server may first perform edge detection on the finger vein image to be identified to obtain the finger edge. The edge detection method may be any one of the Sobel operator, Laplacian operator, Canny operator, etc. The server may locate the finger edge by performing edge detection on the vein image to be identified. In one embodiment, the server may also perform binarization and finger body extraction processing on the finger vein image after edge detection, and then perform midline fitting, and use the fitted midline to correct the erroneous boundary, thereby obtaining the final boundary, thereby improving the accuracy of finger edge detection.
[0064] Since the finger is not fixed during the acquisition process, the acquired finger vein image is not completely horizontal, and most of them have a certain angle offset. The angle between the midline of the finger body and the horizontal direction can be defined as the tilt angle. This tilt will affect the alignment of the region of interest and ultimately lead to the failure of finger vein recognition. Therefore, the server can perform direction correction on the finger vein image. The server can record the coordinates of the edge points in the edge of the finger, calculate the tilt angle of the finger according to the recorded edge point coordinates, and rotate the finger vein image according to the tilt angle. For example, bilinear interpolation can be used to rotate the image. In one embodiment, the server can determine the upper and lower edge points in the recorded edge point coordinates, calculate the midpoint coordinate set through the upper and lower edge points, fit a straight line according to the midpoint coordinate set, and obtain the finger's tilt angle according to the straight line and the calculation formula of the slope and angle.
[0065] Since the finger vein images to be identified collected by the image acquisition device are not for the entire finger, but the main part of the finger, it is difficult to locate the horizontal position of the finger through edge detection. Therefore, in order to solve the influence of horizontal translation on the comparison of homologous images, the server can crop the finger vein images after direction correction. The server determines the vertical interception boundary based on the innermost edge points of the upper and lower edges of the finger edge. With the determined vertical interception reference line as the edge, calculate the grayscale cumulative value of all rows in the area. Find the position corresponding to the maximum grayscale cumulative value, which is the finger joint position. With the finger joint position found as the horizontal reference line, set the preset length before and after to determine the horizontal interception boundary, so that the server can crop the finger vein image after direction correction according to the horizontal interception boundary, and then crop a relatively stable area of interest.
[0066] The cropped finger vein image is no longer disturbed by background information, and the influence of factors such as translation and rotation is basically eliminated. However, the size of the cropped finger vein image is not uniform, which is not conducive to subsequent processing. At the same time, it cannot avoid the impact of scaling caused by the inconsistent distance between the finger and the image acquisition area of the image acquisition device on the comparison of homologous images. Therefore, the server can normalize the cropped finger vein image to obtain a stable target area image with exactly the same size. For example, the normalization process can be based on the preset length and width, and the bilinear interpolation method is used to normalize the cropped finger vein image. For example, the length is 370 pixels and the width is 130 pixels.
[0067] Step 206, calling a pre-built filter, where the filter is built based on the finger vein growth characteristic information.
[0068] The filter is constructed by the finger vein growth characteristic information. For example, the filter can be a two-dimensional filter, such as 2DGabor, which is used to process the target area image and obtain the finger vein features. The shape of the 2DGabor filter is similar to the two-dimensional receptive field of the mammalian visual cortical cell, which can effectively capture the spatial position, frequency, direction and other characteristics of the target area image. The finger vein growth characteristic information refers to the distribution law of the finger vein in frequency and direction obtained by statistics. Since the finger vein pattern grows along the direction of the finger, for example, if the finger pointing is the horizontal direction, the finger vein will be more distributed near the horizontal direction and less distributed in other directions. At the same time, the vein thickness of different finger areas, such as the knuckle area and the finger body area outside the knuckle area, is different. The main vein is thicker, and the branch veins and peripheral veins are thinner. Therefore, the finger vein growth characteristic information can include the distribution law of the finger veins in frequency and direction obtained by statistics in different finger areas.
[0069] By obtaining the finger vein growth characteristic information in advance to construct a filter, the server can call the pre-constructed filter to extract the finger vein features that meet the finger vein growth characteristic information after obtaining the target area image.
[0070] Step 208 , extracting features from the target area image through a filter to obtain finger vein features corresponding to the finger vein growth characteristic information.
[0071] The server can extract the finger vein features in the target area image through the filter. The finger vein features refer to the finger vein features that meet the finger vein growth characteristics, which can be reflected as an amplitude response graph and a frequency direction response graph. The frequency direction response graph refers to a response graph that considers both the frequency parameter and the direction parameter. Specifically, the target area image is divided into multiple units through the filter. The unit can be used to represent the finger area, that is, the target area image is divided into different finger areas. For example, the target area image can be divided into A×B non-overlapping units. For example, since the filter is constructed based on the finger vein growth characteristic information, and the finger vein growth characteristic information includes the distribution law of the finger veins in different finger areas in frequency and direction obtained by statistics. The distribution law of the finger veins in different finger areas in frequency and direction may be different. The filter can obtain the amplitude response graph and the frequency direction response graph of the pixel point that meet the finger vein growth characteristic information for each unit, so as to generate a corresponding histogram according to the amplitude response graph and the frequency direction response graph. And the histogram is output as the finger vein feature.
[0072] Step 210, comparing the finger vein feature with the preset feature to obtain feature similarity.
[0073] Step 212, selecting a preset feature whose feature similarity is greater than a threshold, and determining that the identity information corresponding to the preset feature is a finger vein recognition result.
[0074] The server pre-stores preset features, which refer to the finger vein features of the user who has been identified, and can be used as a finger vein feature template to identify the finger vein features of the finger vein image to be identified. Specifically, the server compares the obtained finger vein features with the preset features to obtain the feature similarity. In one embodiment, the server can use the Euclidean distance method to calculate the feature similarity between the finger vein features and each preset feature. The server compares the obtained multiple feature similarities with a threshold. For example, the threshold can be 95%. When the feature similarity is greater than the threshold, it means that the similarity between the mobile phone vein feature and the preset feature corresponding to the feature similarity is high, then the server can select the preset feature whose similarity is greater than the threshold, and then determine that the identity information corresponding to the preset feature is the finger vein recognition result.
[0075] In this embodiment, a finger vein image to be identified is obtained, and a region of the finger vein image to be identified is intercepted to obtain a target region image, thereby calling a pre-constructed filter, which is constructed based on the finger vein growth characteristic information. The filter is used to extract features of the target region image to obtain the finger vein features corresponding to the finger vein growth characteristic information. Then, the finger vein features are compared with the preset features to obtain feature similarity, and the preset features with feature similarity greater than the threshold are selected to determine the identity information corresponding to the preset features as the finger vein recognition result. Since the filter is constructed based on the finger vein growth characteristic information, the finger vein features can be extracted based on the distribution law of the frequency and direction of the finger veins in different finger regions obtained by statistics, so that the finger vein features are more consistent with the finger vein generation characteristic information. Compared with the traditional general filter that can only extract finger vein features of specific aspects in the finger vein image, more effective and accurate finger vein features can be extracted, thereby improving the accuracy of finger vein recognition. At the same time, the adaptability of the filter to the finger vein patterns in the finger vein features and the extraction efficiency of the finger vein features are also improved.
[0076] In one embodiment, Figure 3 As shown, the steps of extracting features of the target area image through a filter to obtain finger vein features corresponding to the finger vein growth characteristic information include:
[0077] Step 302: Divide the target area image into multiple units through a filter.
[0078] Step 304 , calculating a first amplitude response map and a first frequency direction response map corresponding to the finger vein growth characteristic information according to the pixel points in each unit.
[0079] Step 306 , generating a target histogram according to the first amplitude response graph and the first frequency direction response graph, and determining that the target histogram is a finger vein feature corresponding to the finger vein growth characteristic information.
[0080] The server may divide the target area image into multiple units through a filter. Each unit may include multiple pixels. For example, the target area image may be divided into A×B non-overlapping units. For each unit, the first amplitude response map and the first frequency direction response map corresponding to the finger vein growth characteristic information are calculated according to the pixel points through a filter. The calculation formula may be as follows:
[0081] m B (x,y)=min(F m (x,y)) (1)
[0082] fθ B (x,y)=arg min m (Fm (x,y)) (2)
[0083] Among them, (x, y) represents the coordinates of the pixel point in each unit, m B (x, y) represents the first amplitude response map, F m (x, y) represents the filter response diagram in m frequency directions, fθ B (x, y) represents the first frequency direction response diagram.
[0084] In one embodiment, calculating the first amplitude response map and the first frequency direction response map corresponding to the finger vein growth characteristic information based on the pixel points includes: calculating the filter response map that conforms to the finger vein growth characteristic information corresponding to each pixel point in each unit; calculating the first amplitude response map and the first frequency direction response map corresponding to the finger vein growth characteristic information based on the filter response map. Specifically, for each unit, the filter response map that conforms to the finger vein growth characteristic information corresponding to each pixel point is obtained through a filter. The calculation formula of the filter response map can be as follows:
[0085] F m (x,y)=C(x,y)*G m ,m∈M (3)
[0086] Among them, F m (x, y) represents the filter response map in the m frequency directions, C(x, y) represents the coordinates of each pixel in unit C, * represents the convolution operation, G m represents a filter constructed according to the frequency direction in the finger vein growth characteristic information, and M represents the number of frequency directions in the finger vein growth characteristic information.
[0087] The filter response graph is input into the above formula (1) through a filter to obtain the first amplitude response graph corresponding to the finger vein growth characteristic information, and the filter response graph is input into the above formula (2) to obtain the first frequency direction response graph corresponding to the finger vein growth characteristic information.
[0088] In order to make the extracted finger vein features more robust to brightness changes and a small amount of translation and rotation of the image to be identified, a target histogram can be generated according to the first amplitude response map and the first frequency direction response map through a filter, and then the target histogram can be determined as the finger vein features corresponding to the finger vein growth characteristic information. For example, based on the histogram of oriented gradients (HOG) method, a target histogram can be generated according to the first amplitude response map and the first frequency direction response map through a filter.
[0089] In this embodiment, the target area image is divided into multiple units by a filter, and the first amplitude response map and the first frequency direction response map corresponding to the finger vein growth characteristic information are calculated according to the pixel points in each unit. The target histogram is generated according to the first amplitude response map and the first frequency direction response map, and the target histogram is determined to be the finger vein feature corresponding to the finger vein growth characteristic information. By dividing the target area image into multiple units, it is convenient to generate the target histogram later. By generating the target histogram, the finger vein feature has strong robustness to the brightness change and a small amount of translation and rotation of the image to be identified. The filter can effectively capture the spatial position, frequency, direction and other characteristics of the image. Combining the filter with the method of generating the histogram for feature extraction can extract more accurate finger vein features.
[0090] In one embodiment, Figure 4 As shown, the step of generating a target histogram according to the first amplitude response graph and the first frequency direction response graph, and determining that the target histogram is a finger vein feature corresponding to the finger vein growth characteristic information specifically includes:
[0091] Step 402: Generate a unit histogram according to the first amplitude response graph and the first frequency direction response graph.
[0092] Step 404: Combine adjacent unit histograms according to preset parameters to generate a block histogram.
[0093] Step 406: normalize the block histogram to obtain a standard histogram.
[0094] Step 408, combining the standard histograms to obtain the finger vein features corresponding to the finger vein growth characteristic information.
[0095] The server may generate a unit histogram according to the first amplitude response graph and the first frequency direction response graph through a filter. The calculation formula for generating the unit histogram may be as follows:
[0096] HC(m) i =∑m B (x,y)if(x,y)∈uniti&&fθ B (x,y)∈bin(m) (4)
[0097] Among them, HC(m) i represents the unit histogram corresponding to unit i, m B (x, y) represents the first amplitude response diagram, fθ B (x, y) represents the first frequency direction response diagram.
[0098] Adjacent unit histograms are combined into blocks by filters according to preset parameters. There can be overlap between two adjacent blocks. The preset parameters can be expressed as b 1 ×b 2 Represented by. All unit histograms in each block are connected in series through the filter to generate a block histogram, which can be expressed as HB j ={HC 1 ,HC 2 ,...,HC b1×b2 For example, when the number of units of the target area image division is A×B and the preset parameter is 2×2, adjacent 2×2 units can be formed into a block, and the number of generated block histograms is N=(A-1)(B-1).
[0099] In order to avoid the situation where the divisor is zero, the block histogram can be normalized by using the L2 norm through the filter. The formula for the normalization process can be shown as follows:
[0100]
[0101] Among them, NHB j Represents the standard histogram obtained after normalization, HB j represents block histogram, and ε is a small constant.
[0102] All standard histograms are combined in series through the filter to obtain the finger vein feature corresponding to the finger vein growth characteristic information feature = {NHB 1 ,NHB 2 ,…,NHB N}, where N represents the number of blocks composed of unit histograms, that is, the number of block histograms. The finger vein feature can be expressed in vector form.
[0103] In this embodiment, a unit histogram is generated according to the first amplitude response map and the first frequency direction response map, and adjacent unit histograms are combined according to preset parameters to generate a block histogram, and the block histogram is normalized to obtain a standard histogram. This can avoid the occurrence of a zero divisor and improve the accuracy of subsequent standard histograms. The standard histograms are combined to obtain finger vein features corresponding to the finger vein growth characteristic information. By embodying the finger vein features in the form of a histogram, the finger vein features can be made more robust to brightness changes and a small amount of translation and rotation of the image to be identified.
[0104] In one embodiment, Figure 5 As shown, the above method also includes: a step of constructing a filter, specifically including:
[0105] Step 502: Acquire multiple finger vein sample images.
[0106] Step 504 , performing region capture on the plurality of finger vein images to obtain a plurality of region images to be processed.
[0107] Step 506, calculating the finger vein growth characteristic information based on the pixel points in the plurality of images of the area to be processed.
[0108] Step 508: construct a filter according to the finger vein growth characteristic information.
[0109] Before obtaining the finger vein image to be identified, the server can pre-acquire the finger vein characteristic information to construct a filter to extract more accurate finger vein features and improve the accuracy of finger vein recognition.
[0110] Specifically, the server obtains multiple finger vein sample images. The finger vein sample images may be collected by different image acquisition devices, or directly use existing public data sets. The acquisition background and acquisition specifications of the finger vein sample images may be different. Further, in order to ensure that the acquired finger vein growth characteristic information is universal, multiple finger vein images of multiple people may be obtained as finger vein sample images. The server performs regional interception on each finger vein sample image to obtain an image of the area to be processed corresponding to each finger vein sample image. Each finger vein sample image corresponds to an image of the area to be processed, and finally obtains multiple images of the area to be processed. In the process of constructing the filter, the method of regional interception of the finger vein sample image to be identified in the actual application process may be adopted. Specifically, the server may first perform edge detection on the finger vein sample image to obtain the finger edge, and then perform direction correction on the finger vein sample image after edge detection. Thus, the server performs cropping processing on the finger vein sample image after direction correction to obtain a relatively stable region of interest. Then, the cropped finger vein image is normalized to obtain a stable image of the area to be processed with exactly the same size.
[0111] The server can set multiple directions and frequencies to construct a multi-scale and multi-directional universal filter. The construction of the filter and the calculation of the direction and frequency can be shown as follows:
[0112]
[0113]
[0114]
[0115] in, σ represents the standard deviation of the Gaussian envelope, θ k Indicates the direction of the filter, O indicates the total number of directions set, and f vrepresents the frequency of the filter, f step Represents the interval between different frequency kernel functions, and F represents the total number of set frequencies.
[0116] The server calculates the finger vein growth characteristic information based on the pixel points in the multiple images of the area to be processed. In one embodiment, the finger vein growth characteristic information is calculated based on the pixel points in the multiple images of the area to be processed, including: calculating the second amplitude response map and the second frequency direction response map corresponding to the corresponding image of the area to be processed based on the pixel points in each image of the area to be processed, and obtaining multiple second amplitude response maps and multiple second frequency response maps; and calculating the finger vein growth characteristic information based on the multiple second amplitude response maps and multiple second frequency direction response maps.
[0117] Specifically, for each image of the area to be processed, the server calculates the filter response map corresponding to each pixel point, and calculates the second amplitude response map and the second frequency direction response map according to the filter response map. The calculation formula of the filter response map can be as follows:
[0118] F kv (x, y) = I (x, y) * G kv ,k=1,2,…,O; v=1,2,…,F (9)
[0119] Among them, F kv (x, y) represents the filter response image at frequency v and direction k, I(x, y) represents the coordinates of each pixel in the image I of the area to be processed, * represents the convolution operation, G kv Indicates the filter, O indicates the total number of directions set, and F indicates the total number of frequencies set.
[0120] The calculation formula for calculating the second amplitude response diagram and the second frequency direction response diagram according to the filter response diagram can be shown as follows:
[0121] m(x, y) = min(F kv (x,y)) (10)
[0122] fθ(x, y) = arg min k,v (F kv (x,y)) ( 11 )
[0123] Where (x, y) represents the coordinates of the pixels in each area to be processed, m(x, y) represents the second amplitude response map, and F kv (x, y) represents the filter response diagram at frequency v and direction k, and fθ(x, y) represents the second frequency direction response diagram.
[0124] After calculating and obtaining the second amplitude response map and the second frequency direction response map corresponding to each image of the area to be processed, the server can calculate the finger vein growth characteristic information according to the second amplitude response map and the second frequency direction response map.
[0125] In this embodiment, multiple finger vein images are region-intercepted to obtain multiple region images to be processed, and the second amplitude response map and the second frequency direction response map corresponding to the corresponding region images to be processed are calculated according to the pixel points in each region image to be processed, and the finger vein growth characteristic information is calculated according to the multiple second amplitude response maps and the multiple second frequency direction response maps, and a filter is constructed according to the finger vein growth characteristic information. Since the finger vein growth characteristic information includes the characteristic information of frequency and direction obtained by statistics, the constructed filter can extract the finger vein features that meet the finger vein growth characteristic information, thereby effectively improving the accuracy of finger vein recognition.
[0126] In one embodiment, the finger vein growth characteristic information is calculated based on multiple second amplitude response maps and multiple second frequency direction response maps, including: dividing each image of the area to be processed into multiple units; accumulating multiple second amplitude response maps corresponding to the same unit at a preset frequency and a preset direction to obtain a cumulative histogram corresponding to each unit; selecting a target frequency and a target direction from the cumulative histogram corresponding to each unit to obtain a target frequency and a target direction corresponding to each unit; and generating the finger vein growth characteristic information based on the target frequencies and target directions corresponding to the multiple units.
[0127] In order to obtain the distribution pattern of the finger veins in different finger areas in frequency and direction, the server can divide each image of the area to be processed into multiple units in the same way, and each unit can include multiple pixels. For example, each image of the area to be processed can be divided into a×b non-overlapping units. Since the unit division method of multiple images of the area to be processed is the same, the units obtained by dividing the multiple images of the area to be processed are also the same. The server can accumulate the second amplitude response map corresponding to the same unit according to the preset frequency and preset direction to obtain the cumulative histogram corresponding to each unit. The cumulative histogram corresponding to each unit has a total of O×F bits, where O represents the total number of directions set, and F represents the total number of frequencies set. The cumulative histogram can be calculated according to the following formula:
[0128]
[0129] Among them, HCR i (kv) represents the cumulative histogram corresponding to unit i, S represents the number of finger vein sample images, that is, the number of images of the area to be processed, m(x, y) represents the second amplitude response map, and fθ(x, y) represents the second frequency direction response map.
[0130] The server selects the target frequency and target direction from the cumulative histogram corresponding to each unit. Specifically, the server arranges the cumulative histogram of each unit in descending order from large to small according to the amplitude. The frequency direction corresponding to the smaller amplitude is the frequency direction distribution law of the finger veins of the unit. The divided units can correspond to different finger areas, so the server can obtain the distribution law of the finger veins in different finger areas in terms of frequency and direction. Furthermore, for each unit, the server can select the frequencies and directions of the first M (M<O×F) filters. In the area where the unit is located, the finger veins with the selected frequencies and directions are more distributed. The selected frequency is used as the target frequency, and the selected direction is used as the target direction. For example, the distribution diagram of the finger veins in the target direction can be as follows. Figure 6 As shown, the line composed of white dots indicates the target direction of the finger vein distribution. Therefore, the finger vein growth characteristic information can be generated according to the target frequency and target direction corresponding to the multiple units, and the filter can be set according to the finger vein growth characteristic information, so that the filter is more in line with the growth characteristics of the finger vein, and the finger vein characteristics can be more effectively obtained.
[0131] It should be understood that although Figures 2 to 5 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figures 2 to 5 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.
[0132] In one embodiment, Figure 7 As shown, a finger vein recognition device is provided, including: an image acquisition module 702, a region interception module 704, a calling module 706, a feature extraction module 708, a feature comparison module 710 and a result determination module 712, wherein:
[0133] The image acquisition module 702 is used to acquire the finger vein image to be identified.
[0134] The region interception module 704 is used to intercept the region of the finger vein image to be identified to obtain a target region image.
[0135] The calling module 706 is used to call a pre-built filter, where the filter is built based on the finger vein growth characteristic information.
[0136] The feature extraction module 708 is used to extract features from the target area image through a filter to obtain finger vein features corresponding to the finger vein growth characteristic information.
[0137] The feature comparison module 710 is used to compare the finger vein feature with the preset feature to obtain feature similarity.
[0138] The result determination module 712 is used to select preset features whose feature similarity is greater than a threshold value, and determine the identity information corresponding to the preset features as the finger vein recognition result.
[0139] In one embodiment, the feature extraction module 708 is also used to divide the target area image into multiple units through a filter; calculate the first amplitude response map and the first frequency direction response map corresponding to the finger vein growth characteristic information based on the pixel points in each unit; generate a target histogram based on the first amplitude response map and the first frequency direction response map, and determine that the target histogram is the finger vein feature corresponding to the finger vein growth characteristic information.
[0140] In one embodiment, the feature extraction module 708 is also used to calculate the filter response map corresponding to each pixel point in each unit that conforms to the finger vein growth characteristic information; and calculate the first amplitude response map and the first frequency direction response map corresponding to the finger vein growth characteristic information based on the filter response map.
[0141] In one embodiment, the feature extraction module 708 is also used to generate a unit histogram based on the first amplitude response map and the first frequency direction response map; combine adjacent unit histograms according to preset parameters to generate a block histogram; normalize the block histogram to obtain a standard histogram; combine the standard histograms to obtain finger vein features corresponding to the finger vein growth characteristic information.
[0142] In one embodiment, the above device further comprises:
[0143] The image acquisition module 702 is also used to acquire multiple finger vein sample images.
[0144] The region interception module 704 is further used to perform region interception on the plurality of finger vein images to obtain a plurality of region images to be processed.
[0145] The calculation module is used to calculate the finger vein growth characteristic information based on the pixel points in the multiple images of the area to be processed.
[0146] A building module is used to build a filter based on the finger vein growth characteristic information.
[0147] In one embodiment, the calculation module is also used to calculate the second amplitude response map and the second frequency direction response map corresponding to the corresponding image of the area to be processed according to the pixel points in each image of the area to be processed, and obtain multiple second amplitude response maps and multiple second frequency response maps; the finger vein growth characteristic information is calculated according to the multiple second amplitude response maps and the multiple second frequency direction response maps.
[0148] In one embodiment, the calculation module is also used to divide each image of the area to be processed into multiple units; accumulate multiple second amplitude response maps corresponding to the same unit according to a preset frequency and a preset direction to obtain a cumulative histogram corresponding to each unit; select a target frequency and a target direction from the cumulative histogram corresponding to each unit to obtain a target frequency and a target direction corresponding to each unit; generate finger vein growth characteristic information according to the target frequencies and target directions corresponding to multiple units.
[0149] For the specific definition of the finger vein recognition device, please refer to the definition of the finger vein recognition method above, which will not be repeated here. Each module in the above-mentioned finger vein recognition device can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0150] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data of a finger vein recognition method. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a finger vein recognition method is implemented.
[0151] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0152] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above embodiments when executing the computer program.
[0153] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above embodiments are implemented.
[0154] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0155] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0156] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.
Claims
1. A finger vein recognition method, characterized in that: The method comprises: Obtaining a finger vein image to be identified; Performing regional interception on the finger vein image to be identified to obtain a target area image, including: performing edge detection on the finger vein image to be identified; calculating the inclination angle of the finger according to the coordinates of the edge points in the finger edge in the finger vein image to be identified after edge detection, and rotating the finger vein image to be identified according to the inclination angle to obtain the finger vein image to be identified after direction correction; determining a vertical interception boundary in the finger vein image to be identified after direction correction, taking the determined vertical interception boundary as an edge, calculating the grayscale cumulative values of all rows in the boundary area, and taking the position corresponding to the maximum grayscale cumulative value as the finger joint position; taking the finger joint position to be found as a horizontal reference line, setting a preset length before and after to determine the horizontal interception boundary; cropping the finger vein image after direction correction according to the horizontal interception boundary to obtain the target area image; Calling a pre-built filter, wherein the filter is built based on finger vein growth characteristic information; the finger vein growth characteristic information includes statistically obtained distribution patterns of finger veins in frequency and direction in different finger regions; The method comprises: extracting features of the target area image by the filter to obtain finger vein features corresponding to the finger vein growth characteristic information, including: dividing the target area image into a plurality of units by the filter; the units are used to represent the finger area; obtaining a first amplitude response map and a first frequency direction response map of the pixel points of each unit that conform to the finger vein growth characteristic information; generating a target histogram according to the first amplitude response map and the first frequency direction response map by a gradient directional histogram method, and determining that the target histogram is the finger vein feature corresponding to the finger vein growth characteristic information; Comparing the finger vein feature with the preset feature to obtain feature similarity; A preset feature whose feature similarity is greater than a threshold is selected, and the identity information corresponding to the preset feature is determined as a finger vein recognition result.
2. The method according to claim 1, characterized in that Performing regional interception on the finger vein image to be identified to obtain a target area image includes: Performing edge detection on the finger vein image to be identified; The finger vein image after edge detection is binarized, the finger body is extracted and the midline is fitted. The fitted midline is used to correct the erroneous boundary and obtain the final boundary. The finger vein image to be identified is intercepted according to the final boundary to obtain a target area image.
3. The method according to claim 2, characterized in that The first amplitude response map and the first frequency direction response map corresponding to the finger vein growth characteristic information are calculated according to the pixel points in each unit, including: Calculate the filter response map corresponding to each pixel point in each unit and in accordance with the finger vein growth characteristic information; A first amplitude response graph and a first frequency direction response graph corresponding to the finger vein growth characteristic information are calculated according to the filter response graph.
4. The method according to claim 2, characterized in that: The step of generating a target histogram according to the first amplitude response graph and the first frequency direction response graph, and determining that the target histogram is a finger vein feature corresponding to the finger vein growth characteristic information includes: generating a unit histogram according to the first amplitude response graph and the first frequency direction response graph; Combining adjacent unit histograms according to preset parameters to generate a block histogram; Normalizing the block histogram to obtain a standard histogram; The standard histograms are combined to obtain the finger vein features corresponding to the finger vein growth characteristic information.
5. The method according to claim 1, characterized in that Before acquiring the finger vein image to be identified, the method further includes: Acquire multiple finger vein sample images; Performing region interception on multiple finger vein images to obtain multiple region images to be processed; The finger vein growth characteristic information is obtained by calculating the pixel points in the plurality of images of the area to be processed; A filter is constructed according to the finger vein growth characteristic information.
6. The method according to claim 5, characterized in that The finger vein growth characteristic information is calculated based on the pixel points in the plurality of images of the area to be processed, including: Calculating a second amplitude response map and a second frequency direction response map corresponding to the corresponding image of the region to be processed according to the pixel points in each image of the region to be processed, to obtain a plurality of second amplitude response maps and a plurality of second frequency response maps; The finger vein growth characteristic information is calculated based on the multiple second amplitude response graphs and the multiple second frequency direction response graphs.
7. The method according to claim 6, characterized in that The finger vein growth characteristic information calculated according to the plurality of second amplitude response graphs and the plurality of second frequency direction response graphs comprises: Divide each image of the area to be processed into a plurality of units; Accumulating a plurality of second amplitude response graphs corresponding to the same unit according to a preset frequency and a preset direction to obtain a cumulative histogram corresponding to each unit; Select the target frequency and target direction in the cumulative histogram corresponding to each unit to obtain the target frequency and target direction corresponding to each unit; Finger vein growth characteristic information is generated according to target frequencies and target directions corresponding to the multiple units.
8. A finger vein recognition device, characterized in that: The device comprises: An image acquisition module, used to acquire a finger vein image to be identified; A region interception module is used to perform region interception on the finger vein image to be identified to obtain a target region image, including: performing edge detection on the finger vein image to be identified; calculating the inclination angle of the finger according to the coordinates of the edge points in the finger edge in the finger vein image to be identified after edge detection, and rotating the finger vein image to be identified according to the inclination angle to obtain the direction-corrected finger vein image to be identified; determining a vertical interception boundary in the direction-corrected finger vein image to be identified, taking the determined vertical interception boundary as an edge, calculating the grayscale cumulative values of all rows in the boundary area, and taking the position corresponding to the maximum grayscale cumulative value as the finger joint position; taking the finger joint position to be found as a horizontal reference line, setting a preset length before and after to determine the horizontal interception boundary; cropping the direction-corrected finger vein image according to the horizontal interception boundary to obtain the target region image; A calling module, used to call a pre-built filter, wherein the filter is built based on finger vein growth characteristic information; the finger vein growth characteristic information includes the statistically obtained distribution law of finger veins in frequency and direction in different finger regions; A feature extraction module, used to extract features from the target area image through the filter to obtain finger vein features corresponding to the finger vein growth characteristic information, including: dividing the target area image into a plurality of units through the filter; the unit is used to represent the finger area; obtaining a first amplitude response map and a first frequency direction response map of the pixel points of each unit that conform to the finger vein growth characteristic information; generating a target histogram according to the first amplitude response map and the first frequency direction response map through a gradient directional histogram method, and determining that the target histogram is the finger vein feature corresponding to the finger vein growth characteristic information; A feature comparison module, used to compare the finger vein feature with a preset feature to obtain feature similarity; The result determination module is used to select the preset features whose feature similarity is greater than a threshold value, and determine the identity information corresponding to the preset features as the finger vein recognition result.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
