Fingerprint extraction method based on optical coherence tomography and related device

Abstract

The application discloses a kind of based on optical coherence tomography's fingerprint extraction method and related equipment, comprising: by the mode of optical correlation tomography, the three-dimensional image of fingertip is obtained, according to preset image reconstruction mode, three-dimensional image is reconstructed, obtains N layer two-dimensional subsurface fingerprint image, respectively to each layer subsurface fingerprint image is extracted, and the feature map corresponding to each layer subsurface fingerprint image is obtained, and all feature maps are weighted fusion, and obtain fusion feature map, based on fusion feature map, determine the position map and direction map of each minutia point of fingerprint, and the position map and direction map of each minutia point are as target fingerprint information, improve the quality of extracted fingerprint minutia information.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of biometric identification, and in particular to a fingerprint extraction method based on optical coherence tomography and related equipment. BACKGROUND

[0002] Fingerprint is a physiological characteristic with the advantages of uniqueness, lifelong invariability, easy to take and read, etc., and is widely used in various automatic fingerprint identification systems. Among them, the fingerprint minutia as an important feature of the fingerprint becomes the primary basis for fingerprint identification technology due to its relatively stable and robust performance on fingerprint image brightness, contrast, resolution and global distortion.

[0003] The inventor found in the process of implementing the present application that many existing fingerprint collection systems are based on the texture of the skin on the surface of the fingertip. This method has a great impact on the extraction of fingerprint minutia information in the case of fingerprint deformation, wear, wetness or dryness, etc., and affects the accuracy of fingerprint identification. SUMMARY

[0004] Embodiments of the present application provide a fingerprint extraction method, device, computer equipment and storage medium based on optical coherence tomography to improve the quality of the extracted fingerprint minutia information.

[0005] To solve the above technical problems, the present application provides a fingerprint extraction method based on optical coherence tomography, comprising:

[0006] acquiring a three-dimensional image of a fingertip by optical coherence tomography;

[0007] reconstructing the three-dimensional image according to a preset image reconstruction mode to obtain N two-dimensional subsurface fingerprint images, wherein N is a positive integer greater than 1;

[0008] performing feature extraction on each of the subsurface fingerprint images to obtain a feature map corresponding to each of the subsurface fingerprint images, and performing weighted fusion on all the feature maps to obtain a fused feature map;

[0009] determining a position map and a direction map of each minutia of the fingerprint based on the fused feature map, and taking the position map and the direction map of each minutia as target fingerprint information.

[0010] Optionally, the three-dimensional image is composed of a plurality of ground penetrating radar B-scan images, and the reconstructing the three-dimensional image according to a preset image reconstruction mode to obtain N two-dimensional subsurface fingerprint images comprises:

[0011] Based on an edge detection algorithm, an outer edge of a subcutaneous tissue structure in each ground penetrating radar B-scan image in the three-dimensional image is obtained, and fitting is performed based on the outer edge to obtain a continuous curve corresponding to each ground penetrating radar B-scan image;

[0012] According to the distribution of N preset pixel values, the pixels in the continuous curve are classified to obtain N sets of pixel collections of different categories;

[0013] For the i-th set of pixel collections, the pixel values belonging to the continuous curve of the same ground penetrating radar B-scan image in the pixel collection are summed and averaged to obtain a row vector corresponding to the pixel collection of the category, i is a positive integer less than or equal to N;

[0014] For the i-th set of pixel collections, the row vectors corresponding to the pixel collection of the category of each ground penetrating radar B-scan image are combined to obtain an i-th two-dimensional subsurface fingerprint image.

[0015] Optionally, the combination of the row vectors corresponding to the pixel collection of the category of each ground penetrating radar B-scan image to obtain an i-th two-dimensional subsurface fingerprint image includes:

[0016] The row vectors corresponding to the pixel collection of the category of each ground penetrating radar B-scan image are combined to obtain a combination vector;

[0017] Each vector value in the combination vector is mapped to a pixel value in the range of 0-255 using the following formula to obtain an i-th two-dimensional subsurface fingerprint image:

[0018]

[0019] where I x represents the pixel value of a point represented by the x-th vector value, I y represents the pixel value of the x-th vector value after normalization, I min represents the minimum value of the gray value in the pixel value corresponding to each vector value in the vector combination, I max represents the maximum value of the gray value in the pixel value corresponding to each vector value in the vector combination.

[0020] Optionally, the N convolutional neural networks are used to extract features from N layers of the subsurface fingerprint images in parallel, and each convolutional neural network corresponds to a layer of the subsurface fingerprint image.

[0021] Optionally, all the feature maps are fused by using an adaptive convolution weighting manner to obtain the fused feature map.

[0022] Optionally, the position map and the direction map of each minutia of the fingerprint are determined based on the fused feature map, including:

[0023] The fused feature map is segmented by taking a x b as a segmentation scale and c as a segmentation step to obtain m feature blocks, wherein a, b and c are positive integers.

[0024] Each of the feature blocks is expanded into a one-dimensional feature vector.

[0025] Each of the one-dimensional feature vectors is input into a detail recognition model to obtain the position map and the direction map of each minutia of the fingerprint through the detail recognition model, wherein the detail recognition model includes a minutia classification discriminator, a position discriminator and a direction discriminator.

[0026] Optionally, the position map and the direction map of each minutia of the fingerprint are obtained through the detail recognition model, including:

[0027] The one-dimensional feature vector is input into the minutia classification discriminator to determine whether the feature block corresponding to the one-dimensional feature vector contains a fingerprint minutia.

[0028] If the one-dimensional feature vector corresponding to the feature block contains a fingerprint minutia, the one-dimensional feature vector is input into the position discriminator and the direction discriminator, respectively.

[0029] The position map of the minutia corresponding to the one-dimensional feature vector is obtained through the position discriminator, and the direction map of the minutia corresponding to the one-dimensional feature vector is obtained through the direction discriminator.

[0030] To solve the above technical problems, the embodiment of the present application also provides a fingerprint extraction device based on optical coherence tomography, comprising:

[0031] An image acquisition module is configured to acquire a three-dimensional image of a fingertip by using an optical coherence tomography method.

[0032] An image reconstruction module is configured to reconstruct the three-dimensional image according to a preset image reconstruction method to obtain N two-dimensional subsurface fingerprint images, wherein N is a positive integer greater than 1.

[0033] A feature acquisition module is configured to extract features from each of the subsurface fingerprint images to obtain a feature map corresponding to each of the subsurface fingerprint images, and fuse all the feature maps to obtain a fused feature map.

[0034] An information extraction module is configured to determine a position map and a direction map of each minutia of the fingerprint based on the fused feature map, and take the position map and the direction map of each minutia as target fingerprint information.

[0035] Optionally, the three-dimensional image is composed of a plurality of ground penetrating radar B-scan images, and the image reconstruction module comprises:

[0036] A curve fitting unit is configured to obtain an outer edge of a subcutaneous tissue structure in each ground penetrating radar B-scan image in the three-dimensional image based on an edge detection algorithm, and perform fitting based on the outer edge to obtain a continuous curve corresponding to each ground penetrating radar B-scan image.

[0037] A pixel classification unit is configured to classify pixels in the continuous curve according to a distribution of N preset pixel values to obtain N sets of pixel classification.

[0038] A row vector determination unit is configured to, for the i-th set of pixel classification, sum and average pixel values in the set of pixel classification that belong to a continuous curve of a same ground penetrating radar B-scan image to obtain a row vector of the continuous curve corresponding to the set of pixel classification, where i is a positive integer less than or equal to N.

[0039] A two-dimensional image generation unit is configured to, for the i-th set of pixel classification, combine the row vectors of each ground penetrating radar B-scan image corresponding to the set of pixel classification to obtain an i-th two-dimensional subsurface fingerprint image.

[0040] Optionally, the two-dimensional image generation unit comprises:

[0041] A vector combination subunit is configured to combine the row vectors of each ground penetrating radar B-scan image corresponding to the set of pixel classification to obtain a combination vector.

[0042] A pixel calculation subunit is configured to map a pixel value represented by each vector value in the combination vector to a distribution range of 0-255 using the following formula to obtain an i-th two-dimensional subsurface fingerprint image:

[0043]

[0044] wherein, I x represents a pixel value of a point represented by an x-th vector value, I y represents a normalized pixel value of the x-th vector value, I min represents a minimum value of a gray value in pixel values corresponding to each vector value in the vector combination, and I maxrepresenting a maximum value of a gray value in a pixel value corresponding to each vector value in the vector combination.

[0045] Optionally, the information extraction module comprises:

[0046] a feature map segmentation unit configured to segment the fused feature map to obtain m feature blocks, with a x b as a segmentation scale and c as a segmentation step size, wherein a, b and c are positive integers;

[0047] a feature block conversion unit configured to convert each of the feature blocks into a one-dimensional feature vector;

[0048] a detail recognition unit configured to input each of the one-dimensional feature vectors into a detail recognition model to obtain a position map and a direction map of each minutia of the fingerprint through the detail recognition model, wherein the detail recognition model comprises a minutia classification discriminator, a position discriminator and a direction discriminator.

[0049] Optionally, the detail recognition unit comprises:

[0050] a minutia judgment subunit configured to input the one-dimensional feature vector into the minutia classification discriminator to determine whether the feature block corresponding to the one-dimensional feature vector contains a fingerprint minutia;

[0051] a vector input subunit configured to input the one-dimensional feature vector into the position discriminator and the direction discriminator if the feature block corresponding to the one-dimensional feature vector contains the fingerprint minutia;

[0052] a position map and direction map determination subunit configured to determine the position map of the minutia corresponding to the one-dimensional feature vector through the position discriminator and determine the direction map of the minutia corresponding to the one-dimensional feature vector through the direction discriminator.

[0053] To solve the above technical problems, the embodiment of the present application further provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above fingerprint extraction method based on optical coherence tomography.

[0054] To solve the above technical problems, the embodiment of the present application further provides a computer readable storage medium, which stores a computer program, wherein the computer program is executed by a processor to implement the steps of the above fingerprint extraction method based on optical coherence tomography.

[0055] The fingerprint extraction method and device based on optical coherence tomography, the computer device and the storage medium provided by the embodiment of the application obtain a three-dimensional image of a fingertip in an optical coherence tomography mode, reconstruct the three-dimensional image according to a preset image reconstruction mode, obtain N two-dimensional subsurface fingerprint images, extract features of each layer of the subsurface fingerprint images respectively, obtain feature maps corresponding to each layer of the subsurface fingerprint images, and perform weighted fusion on all the feature maps to obtain a fusion feature map, determine a position map and a direction map of each minutia point of the fingerprint based on the fusion feature map, and take the position map and the direction map of each minutia point as target fingerprint information, thereby improving the quality of the extracted fingerprint minutia point information. BRIEF DESCRIPTION OF DRAWINGS

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the application. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor under the premise of the drawings.

[0057] Figure 1 is an exemplary system architecture diagram to which the present application can be applied;

[0058] Figure 2 is a flowchart of one embodiment of the fingerprint extraction method based on optical coherence tomography of the present application;

[0059] Figure 3 is a specific schematic diagram of the fingerprint extraction method based on optical coherence tomography of the present application;

[0060] Figure 4 is a structural schematic diagram of one embodiment of the fingerprint extraction device based on optical coherence tomography according to the present application;

[0061] Figure 5 is a structural schematic diagram of one embodiment of the computer device according to the present application. DETAILED DESCRIPTION

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs; the terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit the present application; the specification, claims and above description of drawings of the present application, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. The specification, claims and above description of drawings of the present application, the terms "first", "second" and the like are used to distinguish different objects, not to describe a particular order.

[0063] Reference to“an embodiment” herein means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase“in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily all referring to a common set of embodiments, of the application, that are alternatives to all other embodiments. It is explicitly contemplated that embodiments described herein can be combined to include claims directed to combinations of the embodiments.

[0064] The technical solutions in the embodiments of the present application will be apparently and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0065] Please refer to Figure 1 As shown in Figure 1 The system architecture 100 can include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a communication link medium between the terminal devices 101, 102, 103 and the server 105. The network 104 can include various connection types, such as wired, wireless communication links, or optical fiber cables, and the like.

[0066] A user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, and the like.

[0067] The terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 (Moving Picture Experts Group Audio Layer III) players, MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop portable computers, desktop computers, and the like.

[0068] The server 105 can be a server providing various services, such as a background server supporting a page displayed on the terminal devices 101, 102, 103.

[0069] It should be noted that the method for extracting a fingerprint based on optical coherence tomography provided by the embodiments of the present application is executed by a server, and accordingly, the device for extracting a fingerprint based on optical coherence tomography is arranged in the server.

[0070] It should be understood that Figure 1 The number of terminal devices, networks and servers in the system is only illustrative. According to the implementation needs, there can be any number of terminal devices, networks and servers, and the terminal devices 101, 102 and 103 in the embodiments of the present application can correspond to application systems in actual production.

[0071] Please refer to Figure 2 , Figure 2 A fingerprint extraction method based on optical coherence tomography is shown, which is applied to the server in the system Figure 1 The method is described in detail as follows:

[0072] S201: Obtain a three-dimensional image of a fingertip by optical coherence tomography.

[0073] The optical coherence tomography (OCT) is an optical diagnostic technique that can perform non-contact and non-invasive tomographic imaging of microscopic structures of living tissues. It uses the basic principle of a weak coherence light interferometer to detect the backscattering or multiple scattering signals of biological tissues at different depths, and through scanning, two-dimensional or three-dimensional structure images of biological tissues can be obtained.

[0074] S202: Reconstruct the three-dimensional image according to a preset image reconstruction method to obtain N two-dimensional subsurface fingerprint images, wherein N is a positive integer greater than 1.

[0075] The preset image reconstruction method is used to construct a two-dimensional image according to a three-dimensional image, which can be selected according to actual needs.

[0076] As a preferred method, the three-dimensional image in the embodiment is composed of a plurality of ground penetrating radar B-scan images. According to the preset image reconstruction method, the three-dimensional image is reconstructed to obtain N two-dimensional subsurface fingerprint images, which includes:

[0077] Based on an edge detection algorithm, the outer edge of the subcutaneous tissue structure in each ground penetrating radar B-scan image in the three-dimensional image is obtained, and based on the outer edge, a continuous curve corresponding to each ground penetrating radar B-scan image is obtained by fitting;

[0078] According to the distribution of N preset pixel values, the pixels in the continuous curve are classified to obtain N pixel sets of the same category;

[0079] For the i-th pixel set of the same category, the pixel values belonging to the continuous curve of the same ground penetrating radar B-scan image in the pixel set are summed and averaged to obtain a row vector corresponding to the category of the pixel set, and i is a positive integer less than or equal to N.

[0080] For the pixel set of the i-th category, each ground penetrating radar B-scan image is combined in the row vector corresponding to the pixel set of the category to obtain the i-th two-dimensional subsurface fingerprint image.

[0081] The fitting continuous curve mode includes but is not limited to: using an analytical expression to approximate discrete data and least square method, etc.

[0082] The ground penetrating radar B-scan image can provide a structure feature with multiple tissue layers, which includes the horny layer, the active epidermis layer and the dermis layer of the subcutaneous fingerprint of the finger tip, etc.

[0083] In a specific embodiment, N is 3, that is, three pixel value distributions are preset, which are (10, 60), [60, 110] and (110, 170) respectively. The pixel points below 10 pixel value and above 170 pixel value are invalid points. The pixel values between 10 pixel and 60 pixel of the continuous curve are accumulated and averaged to form a row vector. The row vectors corresponding to each ground penetrating radar B-scan image are combined to form the first layer subsurface fingerprint image. The pixel values between 60 pixel and 110 pixel of the continuous curve are accumulated and averaged to form a row vector. The row vectors corresponding to each ground penetrating radar B-scan image are combined to form the second layer subsurface fingerprint image. The pixel values between 110 pixel and 170 pixel of the continuous curve are accumulated and averaged to form a row vector. The row vectors corresponding to each ground penetrating radar B-scan image are combined to form the third layer subsurface fingerprint image.

[0084] Further, for the pixel set of the i-th category, each ground penetrating radar B-scan image is combined in the row vector corresponding to the pixel set of the category to obtain the i-th two-dimensional subsurface fingerprint image, which includes:

[0085] Each ground penetrating radar B-scan image is combined in the row vector corresponding to the pixel set of the category to obtain a combination vector.

[0086] Each vector value in the combination vector is mapped to the distribution range of 0-255 by using the following formula to obtain the i-th two-dimensional subsurface fingerprint image:

[0087]

[0088] where I x represents the pixel value of a point represented by the x-th vector value, I y represents the normalized pixel value of the x-th vector value.min I represents the minimum grayscale value among the pixel values ​​corresponding to the vector values ​​in the vector combination. max This represents the maximum grayscale value among the pixel values ​​corresponding to each vector value in the vector combination.

[0089] S203: Extract features from each subsurface fingerprint image to obtain the feature map corresponding to each subsurface fingerprint image, and then perform weighted fusion of all feature maps to obtain the fused feature map.

[0090] Optionally, N convolutional neural networks are used to extract features from N layers of subsurface fingerprint images in parallel, with each convolutional neural network corresponding to one layer of subsurface fingerprint images.

[0091] Optionally, an adaptive convolutional weighting method is used to weight and fuse all feature maps to obtain a fused feature map.

[0092] Continuing with the example from step S202, as follows: Figure 3 As shown, Figure 3 This diagram illustrates a specific implementation of the fingerprint extraction method based on optical coherence tomography (OCT) of this application. The subsurface fingerprint image of the first layer is input into the first branch of a convolutional neural network (the convolutional neural network corresponding to the subsurface fingerprint image of the first layer). After processing operations such as convolution, nonlinear activation, and pooling, a multi-dimensional feature map fm1(f11, f12, ..., f1n) is obtained, where the dimension of fm1 is h × w × n, h represents the height of the feature map, w represents the width of the feature map, and n represents the number of channels in the feature map. The subsurface fingerprint of the second layer is then input into the convolutional neural network... In the second branch (the convolutional neural network corresponding to the subsurface fingerprint image of the second layer), after processing operations such as convolution, nonlinear activation, and pooling, a multi-dimensional feature map fm2 (f21, f22, ..., f2n) is obtained, with the same dimension as fm1. The subsurface fingerprint of the third layer is input into the third branch of the convolutional neural network (the convolutional neural network corresponding to the subsurface fingerprint image of the third layer), and after processing operations such as convolution, nonlinear activation, and pooling, a multi-dimensional feature map fm3 (f31, f32, ..., f3n) is obtained, with the same dimension as fm1 and fm2. Specifically, the convolutional neural network can be a partial network that removes fully connected layers and some convolutional layers from general backbone networks such as ResNet18 and ResNet50.

[0093] Furthermore, the feature map set {fm1,fm2,fm3} of the subsurface fingerprint is adaptively weighted and fused using a 1×1×3 convolutional kernel k = (k1,k2,k3). The fusion process is as follows:

[0094] fm=fm1×k1+fm2×k2+fm3×k3

[0095] wherein the values k1, k2 and k3 of the convolution kernel represent the weight size of the feature map of the different sub-surface fingerprints to the fused feature map, and the values of the convolution kernel can be further optimized by learning to achieve adaptive fusion; fm represents the fused feature map, i.e., fm = (k1xf11+k2xf21+k3xf31, k1xf12+k2xf22+k3xf32,…, k1xf1n+k2xf2n+k3xf3n).

[0096] S204: Based on the fused feature map, determine the position map and the direction map of each minutia of the fingerprint, and take the position map and the direction map of each minutia as the target fingerprint information.

[0097] Optionally, based on the fused feature map, determining the position map and the direction map of each minutia of the fingerprint comprises:

[0098] Segmenting the fused feature map to obtain m feature blocks, wherein a, b and c are positive integers;

[0099] Each feature block is expanded into a one-dimensional feature vector;

[0100] Each one-dimensional feature vector is input into a detail recognition model to obtain the position map and the direction map of each minutia of the fingerprint through the detail recognition model, wherein the detail recognition model comprises a minutia classification discriminator, a position discriminator and a direction discriminator.

[0101] Continuing with the example in step S203, the segmentation scale is 2x2 and the step is 1. The fused feature map fm of the scale h x w x n after fusion is segmented to obtain (h-1) x (w-1) feature blocks, each of which has a scale of 2 x 2 x n. For the i-th feature block, it is expanded into a one-dimensional feature vector a i (i = 1, 2, …, (h-1) x (w-1)) of size 2 x 2 x n.

[0102] Optionally, the position map and the direction map of each minutia of the fingerprint obtained through the detail recognition model comprise:

[0103] The one-dimensional feature vector is input into the minutia classification discriminator to determine whether the feature block corresponding to the one-dimensional feature vector contains a fingerprint minutia;

[0104] If the result of the determination is that the feature block corresponding to the one-dimensional feature vector contains a fingerprint minutia, the one-dimensional feature vector is input into the position discriminator and the direction discriminator, respectively;

[0105] The position of the minutia corresponding to the one-dimensional feature vector is determined by the position discriminator, and the direction of the minutia corresponding to the one-dimensional feature vector is determined by the direction discriminator.

[0106] Continuing to take the example in step S203 as an example, as shown in the figure, Figure 3 After the feature map is segmented and converted, N one-dimensional feature vectors are obtained, and the one-dimensional feature vector corresponding to each feature block is input into the minutia classification discriminator. The minutia classification discriminator determines whether the feature block corresponding to the one-dimensional feature vector contains a fingerprint minutia. Specifically, the minutia classification discriminator outputs a confidence. If the confidence is greater than or equal to a manually set threshold threshold = 0.5, it is determined that the feature block contains a fingerprint minutia. If the confidence is less than the manually set threshold threshold = 0.5, it is determined that the block does not contain a fingerprint minutia. The feature vectors corresponding to confidence >= threshold are input into the position discriminator and the direction discriminator to output the position and direction of the fingerprint minutia. The feature vectors corresponding to confidence < threshold are not processed. A blank image with the same size as the sub-surface fingerprint is set, and the position and direction of the determined minutia are marked on the blank image. After all the feature vectors are processed, the final predicted position map and direction map of the fingerprint minutia are obtained.

[0107] It should be noted that the threshold value can be set according to actual needs, which is not limited here. As a preferred mode, the threshold value in this embodiment is set to 0.5.

[0108] In this embodiment, the three-dimensional image of the fingertip is obtained by optical correlation tomography. According to a preset image reconstruction mode, the three-dimensional image is reconstructed to obtain N two-dimensional sub-surface fingerprint images. The feature of each sub-surface fingerprint image is extracted to obtain a feature map corresponding to each sub-surface fingerprint image. All feature maps are fused to obtain a fused feature map. Based on the fused feature map, the position map and direction map of each minutia of the fingerprint are determined. The position map and direction map of each minutia are used as target fingerprint information, and the quality of the extracted fingerprint minutia information is improved.

[0109] It should be understood that the sequence numbers in the above embodiments do not mean the order of execution. The execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0110] Figure 4A principle block diagram of an optical coherence tomography based fingerprint extraction device corresponding to the above-mentioned embodiment of the optical coherence tomography based fingerprint extraction method is shown. As shown in Figure 4 the optical coherence tomography based fingerprint extraction device includes an image acquisition module 31, an image reconstruction module 32, a feature acquisition module 33 and an information extraction module 34. Each functional module is described in detail as follows:

[0111] The image acquisition module 31 is configured to acquire a three-dimensional image of a fingertip by optical coherence tomography;

[0112] The image reconstruction module 32 is configured to reconstruct the three-dimensional image according to a preset image reconstruction manner to obtain N two-dimensional subsurface fingerprint images, where N is a positive integer greater than 1;

[0113] The feature acquisition module 33 is configured to respectively extract features from each layer of the subsurface fingerprint images to obtain a feature map corresponding to each layer of the subsurface fingerprint images, and to perform weighted fusion on all the feature maps to obtain a fused feature map;

[0114] The information extraction module 34 is configured to determine a location map and a direction map of each minutia of the fingerprint based on the fused feature map, and to take the location map and the direction map of each minutia as target fingerprint information.

[0115] Optionally, the three-dimensional image is composed of a plurality of ground penetrating radar B-scan images, and the image reconstruction module 32 includes:

[0116] A curve fitting unit is configured to acquire an outer edge of a subcutaneous tissue structure in each ground penetrating radar B-scan image in the three-dimensional image based on an edge detection algorithm, and to perform fitting based on the outer edge to obtain a continuous curve corresponding to each ground penetrating radar B-scan image;

[0117] A pixel classification unit is configured to classify pixels in the continuous curve according to a distribution of N preset pixel values to obtain N sets of pixel collections of different categories;

[0118] A row vector determination unit is configured to, for the i-th set of pixel collections of a category, sum and average pixel values in the set of pixel collections that belong to a continuous curve of the same ground penetrating radar B-scan image to obtain a row vector corresponding to the set of pixel collections of the category, where i is a positive integer less than or equal to N;

[0119] A two-dimensional image generation unit is configured to, for the i-th set of pixel collections of a category, combine the row vectors corresponding to the set of pixel collections of the category in each ground penetrating radar B-scan image to obtain the i-th two-dimensional subsurface fingerprint image.

[0120] Optionally, the two-dimensional image generation unit includes:

[0121] a vector combination subunit, configured to combine each ground penetrating radar B-scan image in a row vector corresponding to a pixel set of a category to obtain a combination vector;

[0122] a pixel calculation subunit, configured to map a pixel value represented by each vector value in the combination vector to a distribution range of 0-255 by using a formula to obtain an i-th two-dimensional subsurface fingerprint image:

[0123]

[0124] wherein, I x represents a pixel value of a point represented by an x-th vector value, I y represents a pixel value after normalization of the pixel value corresponding to the x-th vector value, I min represents a minimum value of a gray value in the pixel value corresponding to each vector value in the vector combination, I max represents a maximum value of a gray value in the pixel value corresponding to each vector value in the vector combination.

[0125] Optionally, the information extraction module 34 comprises:

[0126] a feature map segmentation unit, configured to segment the fused feature map to obtain m feature blocks by taking a×b as a segmentation scale and c as a segmentation step, wherein a, b and c are all positive integers;

[0127] a feature block conversion unit, configured to convert each feature block into a one-dimensional feature vector;

[0128] a detail recognition unit, configured to input each one-dimensional feature vector into a detail recognition model to obtain a position map and a direction map of each minutia of the fingerprint by the detail recognition model, wherein the detail recognition model comprises a minutia classification discriminator, a position discriminator and a direction discriminator.

[0129] Optionally, the detail recognition unit comprises:

[0130] a minutia judgment subunit, configured to input the one-dimensional feature vector into the minutia classification discriminator to judge whether the feature block corresponding to the one-dimensional feature vector contains a fingerprint minutia;

[0131] a vector input subunit, configured to input the one-dimensional feature vector into the position discriminator and the direction discriminator if the feature block corresponding to the one-dimensional feature vector contains the fingerprint minutia according to the judgment result;

[0132] a position map and direction map determination subunit, configured to obtain a position map of the minutia corresponding to the one-dimensional feature vector by the position discriminator and obtain a direction map of the minutia corresponding to the one-dimensional feature vector by the direction discriminator.

[0133] The specific limitations of the fingerprint extraction device based on optical coherence tomography can refer to the limitations of the fingerprint extraction method based on optical coherence tomography described above, which will not be repeated here. Each module in the above fingerprint extraction device based on optical coherence tomography can be realized by software, hardware and their combination. The above modules can be embedded in the processor in the computer device in hardware form or independent of the processor in the computer device, or can be stored in the memory in the computer device in software form, so that the processor calls and executes the operations corresponding to each module.

[0134] To solve the above technical problems, the embodiments of the present application also provide a computer device. For details, please refer to Figure 5 , Figure 5 The basic structure block diagram of the computer device of the present embodiment is shown in FIG. 4.

[0135] The computer device 4 includes a memory 41, a processor 42 and a network interface 43 which are connected to each other through a system bus. It should be noted that only the computer device 4 with the components of the memory 41, the processor 42 and the network interface 43 is shown in the figure, but it should be understood that it is not required to implement all the shown components, and more or less components can be alternatively implemented. Among them, those skilled in the art can understand that the computer device here is a device capable of automatically performing numerical calculation and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessor, application specific integrated circuit (ASIC), field programmable gate array (FPGA), digital signal processor (DSP), embedded device, etc.

[0136] The computer device can be a desktop computer, a notebook computer, a palm computer and a cloud server, etc. The computer device can interact with the user through a keyboard, a mouse, a remote controller, a touchpad or a voice control device, etc.

[0137] The memory 41 includes at least one type of readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or D interface display memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or a memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit and the external storage device of the computer device 4. In the present embodiment, the memory 41 is generally used to store an operating system and various application software installed on the computer device 4, such as program codes for controlling electronic files, etc. In addition, the memory 41 can also be used to temporarily store various data that have been output or will be output.

[0138] The processor 42 can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In the present embodiment, the processor 42 is used to run program codes or process data stored in the memory 41, such as program codes for optical coherence tomography-based fingerprint extraction.

[0139] The network interface 43 can include a wireless network interface or a wired network interface, and is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0140] The present application also provides another embodiment, i.e., to provide a computer readable storage medium storing an interface display program, which can be executed by at least one processor to make the at least one processor perform the steps of the above-described optical coherence tomography-based fingerprint extraction method.

[0141] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and the necessary general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes a plurality of instructions for making a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) execute the method described in each embodiment of the present application.

[0142] Obviously, the above-described embodiments are only some of the embodiments of the present application, not all the embodiments, and the drawings show the preferred embodiments of the present application, but do not limit the patent scope of the present application. The present application can be implemented in many different forms, and conversely, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing specific embodiments, or make equivalent replacements to some of the technical features. Any equivalent structure made by using the content of the specification and drawings, directly or indirectly applied to other related technical fields, is also within the scope of the patent protection of the present application.

Claims

1. An optical coherence tomography based fingerprint extraction method, characterized by, The method comprises the following steps: acquiring a three-dimensional image of a fingertip by means of optical coherence tomography; reconstructing the three-dimensional image according to a preset image reconstruction mode to obtain N two-dimensional subsurface fingerprint images, wherein N is a positive integer greater than 1; extracting features from each of the subsurface fingerprint images to obtain a feature map corresponding to each of the subsurface fingerprint images, and fusing all the feature maps to obtain a fused feature map; determining a location map and a direction map of each minutia of a fingerprint based on the fused feature map, and taking the location map and the direction map of each minutia as target fingerprint information; wherein the three-dimensional image is composed of a plurality of ground penetrating radar B-scan images, and the three-dimensional image is reconstructed according to the preset image reconstruction mode to obtain N two-dimensional subsurface fingerprint images, which comprises: obtaining the outer edge of the subcutaneous tissue structure in each ground penetrating radar B-scan image in the three-dimensional image based on an edge detection algorithm, and fitting based on the outer edge to obtain a continuous curve corresponding to each ground penetrating radar B-scan image; classifying the pixels in the continuous curve according to the distribution of N preset pixel values to obtain N sets of pixel collections of different categories; for the i-th set of pixel collections, summing and averaging the pixel values of the pixels belonging to the continuous curve of the same ground penetrating radar B-scan image in the set of pixel collections to obtain a row vector corresponding to the set of pixel collections of the continuous curve, i being a positive integer less than or equal to N; for the i-th set of pixel collections, combining the row vectors corresponding to the set of pixel collections of each ground penetrating radar B-scan image to obtain the i-th two-dimensional subsurface fingerprint image.

2. The optical coherence tomography-based fingerprint extraction method according to claim 1, wherein, The combining of the row vectors corresponding to the set of pixel collections of each ground penetrating radar B-scan image to obtain the i-th two-dimensional subsurface fingerprint image comprises: combining the row vectors corresponding to the set of pixel collections of each ground penetrating radar B-scan image to obtain a combination vector; mapping the pixel values represented by each vector value in the combination vector to a distribution range of 0-255 using the following formula to obtain the i-th two-dimensional subsurface fingerprint image: wherein represents a pixel value of a point represented by the xth vector value, represents a pixel value of a point represented by the xth vector value, represents a minimum value of a gray value among pixel values corresponding to each vector value in the vector combination, represents a maximum value of a gray value among pixel values corresponding to each vector value in the vector combination.

3. The optical coherence tomography-based fingerprint extraction method of claim 1, wherein, using N convolutional neural networks to extract features from N two-dimensional subsurface fingerprint images in parallel, each convolutional neural network corresponding to one of the two-dimensional subsurface fingerprint images.

4. The optical coherence tomography-based fingerprint extraction method of claim 1, wherein, using an adaptive convolution weighting method to fuse all the feature maps to obtain the fused feature map.

5. The optical coherence tomography based fingerprint extraction method according to any one of claims 1 to 4, wherein, The determination of the location map and the direction map of each minutia of the fingerprint based on the fused feature map comprises: segmenting the fused feature map with a×b as the segmentation scale and c as the segmentation step to obtain m feature blocks, wherein a, b and c are positive integers; expanding each feature block into a one-dimensional feature vector; input each of the one-dimensional feature vectors into a detail recognition model, and recognize, by the detail recognition model, a position map and a direction map of each minutia point of the fingerprint, wherein the detail recognition model comprises a minutia point classification discriminator, a position discriminator, and a direction discriminator.

6. The optical coherence tomography-based fingerprint extraction method according to claim 5, wherein, The recognizing, by the detail recognition model, the position map and the direction map of each minutia point of the fingerprint comprises: inputting the one-dimensional feature vector into the minutia point classification discriminator to determine whether the feature block corresponding to the one-dimensional feature vector contains a fingerprint minutia point; if the determination result is that the feature block corresponding to the one-dimensional feature vector contains a fingerprint minutia point, inputting the one-dimensional feature vector into the position discriminator and the direction discriminator respectively; determining, by the position discriminator, the position map of the minutia point corresponding to the one-dimensional feature vector, and determining, by the direction discriminator, the direction map of the minutia point corresponding to the one-dimensional feature vector.

7. An optical coherence tomography based fingerprint extraction apparatus, characterized by, comprise: an image acquisition module configured to acquire a three-dimensional image of a fingertip by optical coherence tomography; an image reconstruction module configured to reconstruct the three-dimensional image according to a preset image reconstruction manner to obtain N two-dimensional subsurface fingerprint images, wherein N is a positive integer greater than 1; a feature acquisition module configured to perform feature extraction on each of the subsurface fingerprint images to obtain a feature map corresponding to each of the subsurface fingerprint images, and perform weighted fusion on all the feature maps to obtain a fused feature map; an information extraction module configured to determine a position map and a direction map of each minutia point of a fingerprint based on the fused feature map, and take the position map and the direction map of each minutia point as target fingerprint information. The three-dimensional image is composed of a plurality of ground penetrating radar B-scan images, and the image reconstruction module comprises: a curve fitting unit configured to acquire an outer edge of a subcutaneous tissue structure in each ground penetrating radar B-scan image in the three-dimensional image based on an edge detection algorithm, and perform fitting based on the outer edge to obtain a continuous curve corresponding to each ground penetrating radar B-scan image; a pixel classification unit configured to classify pixels in the continuous curve according to the distribution of N preset pixel values to obtain N sets of pixel classification; a row vector determination unit configured to, for the i-th set of pixel classification, sum and average pixel values belonging to a continuous curve of the same ground penetrating radar B-scan image in the set of pixel classification to obtain a row vector corresponding to the set of pixel classification of the continuous curve, wherein i is a positive integer less than or equal to N; a two-dimensional image generation unit configured to, for the i-th set of pixel classification, combine the row vectors corresponding to the set of pixel classification of each ground penetrating radar B-scan image to obtain an i-th two-dimensional subsurface fingerprint image.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the fingerprint extraction method based on optical coherence tomography according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 8. The computer program, when executed by the processor, implements the optical coherence tomography based fingerprint extraction method according to any one of claims 1 to 6.

Images