Non-contact fingerprint image enhancement method and device, non-contact fingerprint image authentication method and device, medium and equipment

Abstract

The embodiment of the invention discloses a non-contact fingerprint image enhancement method and device, an identity verification method and device, a medium and equipment, and the method comprises the steps: firstly inputting a non-contact fingerprint photographing image into a pre-trained fingerprint ridge enhancement model which comprises a fingerprint ridge reconstruction network and a fingerprint image reconstruction network; reconstructing a fingerprint ridge line in the non-contact fingerprint photographing image through a fingerprint ridge line reconstruction network to obtain reconstructed fingerprint ridge line features; reconstructing the non-contact fingerprint photographing image through a fingerprint image reconstruction network to obtain a non-contact fingerprint reconstruction image; and fusing the reconstructed fingerprint ridge line features into the non-contact fingerprint reconstructed image to obtain a non-contact fingerprint enhanced image. According to the technical scheme, the accuracy and robustness of the non-contact fingerprint enhanced image can be effectively improved, the adaptability of the algorithm to non-contact fingerprint photographed images in various photographing scenes is improved, and the quality of the non-contact fingerprint enhanced image is ensured.

Description

Technical Field

[0001] The present invention relates to the field of identity authentication technology, and in particular to a contactless fingerprint image enhancement method, a contactless fingerprint identity authentication method and apparatus, medium, and equipment. Background Art

[0002] With the rapid development of identity authentication technology, fingerprint recognition, as an important part of the field of biometric identification, is widely used in scenarios such as identity authentication, mobile payment, and secure access control. Contactless Fingerprint Identification refers to the process of obtaining and authenticating personal fingerprint information in a contactless manner. Identity authentication usually requires verifying the authenticity of a person's identity. Traditional fingerprint authentication often requires an individual to place a fingerprint on a contact sensor, while contactless fingerprint identification technology can capture fingerprint information remotely without physical contact. Contactless fingerprint identification technology can be used in a variety of ways, such as using a high-resolution camera to capture fingerprint images, or using advanced image processing software algorithms to analyze captured fingerprint features from a distance.

[0003] At present, in order to improve the quality of contactless fingerprint images, image enhancement methods based on traditional image processing technology or deep learning methods are usually adopted in related technologies. However, the fingerprint features represented by the enhanced images obtained by these solutions have low accuracy and poor robustness. Summary of the invention

[0004] In a first aspect of the embodiments of the specification, a contactless fingerprint image enhancement method is provided, which can improve the accuracy and robustness of the output contactless fingerprint enhanced image and ensure the quality of the contactless fingerprint enhanced image. The method includes:

[0005] Acquire a contactless fingerprint photographic image, and input the contactless fingerprint photographic image into a pre-trained fingerprint ridge enhancement model, wherein the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network;

[0006] Reconstructing the fingerprint ridges in the contactless fingerprint photographed image by the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features;

[0007] Reconstructing the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image;

[0008] The reconstructed fingerprint ridge features are fused to the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image.

[0009] Further, in some embodiments, reconstructing the fingerprint ridges in the contactless fingerprint photographed image by the fingerprint ridge reconstruction network to obtain the reconstructed fingerprint ridge features includes:

[0010] Extracting original fingerprint ridge features from the contactless fingerprint photographed image;

[0011] The original fingerprint ridge features are input into the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features.

[0012] Furthermore, in some embodiments, the extracting the original fingerprint ridge features in the contactless fingerprint photographed image includes:

[0013] Extracting high-frequency component data from the contactless fingerprint photographed image by using a high-pass filter based on a Gaussian kernel;

[0014] The high frequency component data is used as the original fingerprint ridge features corresponding to the contactless fingerprint photographed image.

[0015] Further, in some embodiments, the fingerprint ridge reconstruction network includes a feature encoder and a frequency domain decoder, and the inputting the original fingerprint ridge feature into the fingerprint ridge reconstruction network to obtain the reconstructed fingerprint ridge feature includes:

[0016] Inputting the original fingerprint ridge features into the feature encoder to extract ridge structure representation data corresponding to the contactless fingerprint photographed image;

[0017] The ridge structure characterization data is input into the frequency domain decoder and decoded to obtain the reconstructed fingerprint ridge features.

[0018] Further, in some embodiments, the fingerprint image reconstruction network includes a feature encoder and an image decoder shared with the fingerprint ridge reconstruction network; the step of reconstructing the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain the contactless fingerprint reconstructed image includes:

[0019] Inputting the original fingerprint ridge features extracted from the contactless fingerprint photographed image into the feature encoder to extract ridge structure representation data;

[0020] The ridge structure characterization data is input into the image decoder to reconstruct a contactless fingerprint reconstruction image corresponding to the contactless fingerprint photographed image.

[0021] Further, in some embodiments, the step of fusing the reconstructed fingerprint ridge features into the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image includes:

[0022] The reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image are weightedly fused point by point using a dynamic weighting method to obtain a contactless fingerprint enhanced image;

[0023] The dynamic weighting method is a method of allocating weights based on the similarity between the reconstructed fingerprint ridge features and the local features of the contactless fingerprint reconstructed image.

[0024] Furthermore, in some embodiments, the method further comprises:

[0025] Acquire a contactless fingerprint sample image, and randomly degrade the contactless fingerprint sample image to obtain a contactless fingerprint simulation image;

[0026] The pre-constructed fingerprint ridge enhancement model is trained based on the contactless fingerprint sample image and the contactless fingerprint simulation image to obtain a trained fingerprint ridge enhancement model.

[0027] Furthermore, in some implementations, the randomly degrading the contactless fingerprint sample image to obtain a contactless fingerprint simulation image includes:

[0028] Acquire a preset degradation factor library, and determine a plurality of random degradation factors based on the degradation factor library, wherein the random degradation factors include a degradation mode and a random degradation degree corresponding to the degradation mode;

[0029] The contactless fingerprint sample image is degraded by the random degradation factor to obtain a contactless fingerprint simulation image that simulates the contactless fingerprint captured in a real scene.

[0030] Further, in some embodiments, the pre-constructed fingerprint ridge enhancement model includes a preliminarily trained fingerprint ridge reconstruction network and a fingerprint image reconstruction network to be trained;

[0031] The method of training the pre-built fingerprint ridge enhancement model based on the contactless fingerprint sample image and the contactless fingerprint simulation image to obtain a trained fingerprint ridge enhancement model includes:

[0032] Extracting original fingerprint ridge features corresponding to the contactless fingerprint sample image, and extracting simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image;

[0033] Inputting the simulated fingerprint ridge features into a preliminarily trained fingerprint ridge reconstruction network to obtain reconstructed simulated ridge features, and constructing a ridge consistency loss function based on the reconstructed simulated ridge features and the original fingerprint ridge features;

[0034] Inputting the simulated fingerprint ridge features into a fingerprint image reconstruction network to be trained to obtain a reconstructed simulated fingerprint image, and constructing an image consistency loss function based on the reconstructed simulated fingerprint image and the contactless fingerprint sample image;

[0035] Constructing a cycle consistency loss function based on the reconstructed simulated ridge features and the reconstructed simulated fingerprint image;

[0036] The overall training loss function of the fingerprint ridge enhancement model is constructed by the ridge consistency loss function, the image consistency loss function and the cycle consistency loss function, and the pre-constructed fingerprint ridge enhancement model is trained based on the overall training loss function to obtain a trained fingerprint ridge enhancement model.

[0037] Furthermore, in some embodiments, the method further comprises:

[0038] Extracting original fingerprint ridge features corresponding to the contactless fingerprint sample image, and extracting simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image;

[0039] The original fingerprint ridge features are used as self-supervisory signals and the simulated fingerprint ridge features are used as inputs to perform self-supervisory training on the pre-constructed fingerprint ridge reconstruction network to obtain a preliminarily trained fingerprint ridge reconstruction network.

[0040] In a second aspect of the embodiments of the specification, a contactless fingerprint identity authentication method is also proposed, comprising:

[0041] Acquire a contactless fingerprint photographed image, and input the contactless fingerprint photographed image into a pre-trained fingerprint ridge enhancement model to obtain a contactless fingerprint enhanced image;

[0042] Based on the comparison and matching of the contactless fingerprint enhanced image in a preset identity identification database, an identity authentication result of the contactless fingerprint photographed image is obtained.

[0043] In a third aspect of the embodiments of the specification, a contactless fingerprint image enhancement device is also proposed, comprising:

[0044] A fingerprint image input module, used to obtain a contactless fingerprint photographic image, and input the contactless fingerprint photographic image into a pre-trained fingerprint ridge enhancement model, wherein the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network;

[0045] A fingerprint ridge reconstruction module, used to reconstruct the fingerprint ridges in the contactless fingerprint photographed image through the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features;

[0046] A fingerprint image reconstruction module, used to reconstruct the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image;

[0047] The modal feature fusion module is used to fuse the reconstructed fingerprint ridge features into the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image.

[0048] In a fourth aspect of the embodiments of the specification, a contactless fingerprint image enhancement device is also proposed, comprising:

[0049] A fingerprint image enhancement module is used to obtain a contactless fingerprint photographic image and input the contactless fingerprint photographic image into a pre-trained fingerprint ridge enhancement model to obtain a contactless fingerprint enhanced image;

[0050] The fingerprint image verification module is used to compare and match the contactless fingerprint enhanced image in a preset identity identification database to obtain the identity verification result of the contactless fingerprint photographed image.

[0051] According to a fifth aspect of the embodiments of this specification, a computer program product is provided. The computer program product stores at least one instruction, and the at least one instruction is suitable for being loaded by a processor and executing the method steps in the first aspect or the second aspect.

[0052] In a sixth aspect of the embodiments of this specification, a storage medium is further provided, wherein the storage medium stores a computer program, and the computer program is suitable for being loaded by a processor and executing the steps of the method in the first aspect or the second aspect.

[0053] According to a seventh aspect of the embodiments of this specification, there is also provided an electronic device, comprising: a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the steps of the method in the first aspect or the second aspect.

[0054] In the embodiments of the present specification, the contactless fingerprint photographed image is input into a pre-trained fingerprint ridge enhancement model, the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network, and the fingerprint ridges in the contactless fingerprint photographed image are reconstructed by the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features; the contactless fingerprint photographed image is reconstructed by the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image; the reconstructed fingerprint ridge features are fused into the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image. On the one hand, the fingerprint ridges in the contactless fingerprint camera image are reconstructed through the fingerprint ridge reconstruction network included in the fingerprint ridge enhancement model, and the contactless fingerprint camera image is enhanced by combining the reconstructed fingerprint ridge features obtained by reconstruction, which can improve the quality of key fingerprint features in the contactless fingerprint camera image in a targeted manner, thereby improving the accuracy of the contactless fingerprint enhanced image; on the other hand, the reconstructed fingerprint ridge features obtained based on the reconstruction are fused with the contactless fingerprint reconstructed image to obtain the final contactless fingerprint enhanced image, which can effectively improve the quality and clarity of the key fingerprint features in the contactless fingerprint enhanced image, namely the fingerprint ridges. Even if the quality of the input contactless fingerprint camera image is poor, the key features in the contactless fingerprint camera image can be enhanced, thereby improving the robustness of the algorithm in various shooting scenarios; on the other hand, the fingerprint ridge enhancement model can realize the enhanced processing of the contactless fingerprint camera image through a single round of reasoning, with a small amount of calculation, low latency, and saving of computing resources. It can be run on mobile devices with low computing power, and can meet the real-time requirements and portability requirements of identity authentication product reasoning. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 A schematic diagram showing a system architecture of an exemplary application environment in which a contactless fingerprint image enhancement method and device according to an embodiment of the present specification can be applied.

[0056] Figure 2 A schematic diagram of a flow chart of a contactless fingerprint image enhancement method provided in an embodiment of this specification.

[0057] Figure 3 A schematic diagram of a process of reconstructing and restoring fingerprint ridge features provided in an embodiment of this specification.

[0058] Figure 4 A schematic diagram of a process of reconstructing and restoring a contactless fingerprint reconstructed image provided in an embodiment of this specification.

[0059] Figure 5 A schematic diagram of a process for pre-training a fingerprint ridge enhancement model provided in an embodiment of this specification;

[0060] Figure 6A schematic diagram of implementing degradation processing of contactless fingerprint photography images through a degradation factor library provided in an embodiment of this specification.

[0061] Figure 7 A schematic diagram of the principle of enhancing a contactless fingerprint photographic image by using a fingerprint ridge enhancement model provided in an embodiment of this specification.

[0062] Figure 8 A schematic diagram of a process of obtaining a fingerprint ridge reconstruction network through self-supervised learning training provided in an embodiment of this specification.

[0063] Fig. 9 A schematic diagram of the principle of self-supervised learning of a fingerprint ridge reconstruction network provided in an embodiment of this specification.

[0064] Fig.10 A schematic diagram of a flow chart of a contactless fingerprint identity authentication method provided in an embodiment of this specification.

[0065] Fig.11 A schematic diagram of the structure of a contactless fingerprint image enhancement device provided in an embodiment of this specification.

[0066] Fig.12 A schematic diagram of the structure of a contactless fingerprint identity authentication device provided in an embodiment of this specification.

[0067] Fig.13 A schematic diagram of the structure of an electronic device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solutions and advantages of this specification more clear, the technical solutions of this specification will be clearly and completely described below in combination with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.

[0069] Figure 1 A schematic diagram of a system architecture of an exemplary application environment in which a contactless fingerprint image enhancement method or a contactless fingerprint identity authentication method and device according to an embodiment of the present specification can be applied is shown.

[0070] like Figure 1As shown, the system architecture 100 may include one or more of terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc. The terminal devices 101, 102, 103 may be various electronic devices with artificial intelligence (AI) computing capabilities, including but not limited to desktop computers, portable computers, smart phones, tablet computers, etc. It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. According to the implementation requirements, there may be any number of terminal devices, networks and servers. For example, the server 105 may be a server cluster composed of multiple servers.

[0071] The contactless fingerprint image enhancement method or contactless fingerprint identity authentication method provided in the embodiments of this specification is generally performed by the terminal devices 101, 102, and 103, and accordingly, the contactless fingerprint image enhancement device or the contactless fingerprint identity authentication device is generally set in the terminal devices 101, 102, and 103. However, it is easy for those skilled in the art to understand that the contactless fingerprint image enhancement method or the contactless fingerprint identity authentication method provided in the embodiments of this specification can also be performed by the server 105, and accordingly, the contactless fingerprint image enhancement device or the contactless fingerprint identity authentication device can also be set in the server 105, which is not particularly limited in this exemplary embodiment.

[0072] See also Figure 2 , which is a flow chart of a contactless fingerprint image enhancement method provided in the embodiment of this specification. In the embodiment of this specification, the contactless fingerprint image enhancement method can be applied to a terminal device or a server, and this exemplary embodiment does not make any special restrictions on this. Figure 2 The process shown in the figure is described in detail. The contactless fingerprint image enhancement method in the embodiment of this specification may specifically include the following steps:

[0073] Step S210, acquiring a contactless fingerprint photographed image, and inputting the contactless fingerprint photographed image into a pre-trained fingerprint ridge enhancement model, wherein the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network;

[0074] Step S220, reconstructing the fingerprint ridges in the contactless fingerprint photographed image through the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features;

[0075] Step S230, reconstructing the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image;

[0076] Step S240: fusing the reconstructed fingerprint ridge features into the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image.

[0077] According to the contactless fingerprint image enhancement method in the embodiment of the present specification, on the one hand, the fingerprint ridges in the contactless fingerprint photographed image are reconstructed by the fingerprint ridge reconstruction network included in the fingerprint ridge enhancement model, and the contactless fingerprint photographed image is enhanced by combining the reconstructed fingerprint ridge features obtained by reconstruction, which can improve the quality of key fingerprint features in the contactless fingerprint photographed image in a targeted manner, thereby improving the accuracy of the contactless fingerprint enhanced image; on the other hand, the reconstructed fingerprint ridge features obtained by reconstruction are fused with the contactless fingerprint reconstructed image to obtain the final contactless fingerprint enhanced image, which can effectively improve the quality and clarity of the key fingerprint features in the contactless fingerprint enhanced image, namely the fingerprint ridges. Even if the quality of the input contactless fingerprint photographed image is poor, the key features in the contactless fingerprint photographed image can be enhanced, thereby improving the robustness of the algorithm in various shooting scenarios; on the other hand, the fingerprint ridge enhancement model can realize the enhanced processing of the contactless fingerprint photographed image through a single round of reasoning, with a small amount of calculation, low latency, and saving of computing resources. It can be run on mobile devices with low computing power, and can meet the real-time requirements and portability requirements of identity authentication product reasoning.

[0078] Below Figure 2 The contactless fingerprint image enhancement method in is described in detail.

[0079] In step S210, a contactless fingerprint photographed image is acquired, and the contactless fingerprint photographed image is input into a pre-trained fingerprint ridge enhancement model, wherein the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network.

[0080] In an example embodiment of the present specification, a contactless fingerprint photographed image refers to a fingerprint image of a finger collected remotely in a non-contact manner. For example, the contactless fingerprint photographed image can be a fingerprint image directly photographed by a camera on a mobile device at a certain distance from the finger to be collected, or it can be a fingerprint image obtained by photographing the finger to be collected by an external camera connected to a computing device for communication. This example embodiment does not specifically limit the method for obtaining the contactless fingerprint photographed image.

[0081] The contactless fingerprint camera image is an optical color image directly captured by an optical sensor, which is different from the non-optical fingerprint images currently obtained by scanning simulation using contact fingerprint collection sensors. For example, the capacitive fingerprint sensor senses the capacitance change of each capacitor unit based on the sensor array, and converts the capacitance change into a grayscale value to obtain a fingerprint image; and the under-screen fingerprint sensor obtains the fingerprint image based on the optical reflected light intensity distribution or the ultrasonic reflection intensity conversion.

[0082] The fingerprint ridge enhancement model refers to an artificial intelligence model that is pre-trained to reconstruct and restore the fingerprint ridges in the contactless fingerprint camera image so as to enhance the key identity authentication features in the contactless fingerprint camera image. For example, the fingerprint ridge enhancement model can be a neural network model based on an encoder-decoder network structure, or a neural network model based on a generative adversarial network structure. This example embodiment does not specifically limit the type of artificial intelligence model for reconstructing and restoring the contactless fingerprint camera image.

[0083] The fingerprint ridge enhancement model may include a fingerprint ridge reconstruction network and a fingerprint image reconstruction network. The fingerprint ridge reconstruction network is mainly used to reconstruct and restore the fingerprint ridge features in the contactless fingerprint camera image, thereby improving the clarity of the fingerprint ridge features in the contactless fingerprint camera image and ensuring that the contactless fingerprint camera image can be accurately identified in downstream tasks; the fingerprint image reconstruction network is mainly used to reconstruct and restore the global image features of the contactless fingerprint camera image, thereby improving the clarity of the structural representation of the contactless fingerprint camera image taken in various scenarios. The fingerprint ridge enhancement model can effectively ensure the restoration of the image texture details of the contactless fingerprint camera image while effectively ensuring the enhancement of key identification features, improving the accuracy and clarity of the fingerprint structure representation of the enhanced contactless fingerprint camera image, and improving the robustness of the fingerprint ridge enhancement model through the collaborative work of the fingerprint ridge reconstruction network and the fingerprint image reconstruction network, and reconstructing and restoring the fingerprint ridge features and the global image features respectively.

[0084] It can be understood that before the contactless fingerprint photographed image is input into the pre-trained fingerprint ridge enhancement model, the contactless fingerprint photographed image can be preprocessed or normalized. For example, the preprocessing of the contactless fingerprint photographed image may include image background segmentation (such as removing the background in the contactless fingerprint photographed image based on edge detection, threshold segmentation and other technologies to reduce redundant image information), noise removal (such as eliminating or smoothing the noise in the contactless fingerprint photographed image through mean filtering, Gaussian filtering, median filtering, etc., retaining texture details and improving image texture clarity), contrast enhancement, etc. This embodiment does not specifically limit the preprocessing method; normalization processing can be a processing method for adjusting images under different acquisition conditions to a standard input format during model training. For example, normalization processing can include image size adjustment, pixel value standardization and color channel correction, etc. This embodiment does not specifically limit the normalization processing method.

[0085] In step S220, the fingerprint ridges in the contactless fingerprint photographed image are reconstructed by the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features.

[0086] In an example embodiment of the present specification, the fingerprint ridge reconstruction network is mainly used to reconstruct and restore the fingerprint ridge features in the contactless fingerprint camera image, thereby improving the clarity of the fingerprint ridge features in the contactless fingerprint camera image and ensuring that the contactless fingerprint camera image can be accurately identified in downstream tasks.

[0087] Reconstructing fingerprint ridge features refers to reconstructing and restoring the original fingerprint baseline features in the contactless fingerprint camera image, focusing on the blurred and broken ridge lines caused by the collection environment or equipment problems. For example, the fingerprint ridge reconstruction network can be composed of a feature encoder and a frequency domain decoder. The feature encoder can extract the structural feature representation in the original ridge features, such as the ridge geometry, ridge direction distribution and other characterization information, and input it into the frequency domain decoder. The frequency domain decoder reconstructs and restores the original ridge features to generate reconstructed fingerprint ridge features with higher texture details, clarity and coherence.

[0088] In step S230, the contactless fingerprint photographed image is reconstructed by the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image.

[0089] In an example embodiment of the present specification, the fingerprint image reconstruction network is mainly used to optimize the overall quality of the contactless fingerprint camera image from the perspective of the global image. For example, the fingerprint image reconstruction network may include a feature encoder and an image decoder. Its working principle is to utilize the spatial distribution information of the image features, extract the global characterization data through the feature encoder, and restore the contactless fingerprint reconstructed image with balanced illumination, optimized contrast and clear details through the image decoder, and cooperate with the ridge reconstruction network to achieve the comprehensive optimization of local details and global consistency.

[0090] In step S240, the reconstructed fingerprint ridge features are fused to the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image.

[0091] In an exemplary embodiment of the present specification, the contactless fingerprint enhanced image refers to the enhanced fingerprint image obtained by fusing the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image. The contactless fingerprint enhanced image strengthens the unity of the fingerprint ridge features and the overall image quality by fusing the reconstructed fingerprint ridge features into the contactless fingerprint reconstructed image, and effectively ensures the accuracy and clarity of the contactless fingerprint enhanced image. The fusion of the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image can be achieved by global static weighted fusion, such as by presetting the fusion weights of the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image, and by fusion weights to achieve the fusion of the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image; the fusion of the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image can also be achieved by local dynamic weighted fusion, such as by calculating the weight distribution of the ridge features and the reconstructed image at each pixel point according to the similarity of local features, specifically by constructing a local sliding window for feature comparison, and adjusting the fusion ratio of each position based on a similarity index (such as cosine similarity or Euclidean distance). This exemplary embodiment does not specifically limit the fusion method of the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image.

[0092] The fingerprint ridges in the contactless fingerprint camera image are reconstructed through the fingerprint ridge reconstruction network included in the fingerprint ridge enhancement model, and the contactless fingerprint camera image is enhanced by combining the reconstructed fingerprint ridge features, which can improve the quality of key fingerprint features in the contactless fingerprint camera image in a targeted manner, thereby improving the accuracy of the contactless fingerprint enhanced image; the reconstructed fingerprint ridge features obtained based on the reconstruction are fused with the contactless fingerprint reconstructed image to obtain the final contactless fingerprint enhanced image, which can effectively improve the quality and clarity of the key fingerprint features in the contactless fingerprint enhanced image, namely the fingerprint ridges. Even if the quality of the input contactless fingerprint camera image is poor, the key features in the contactless fingerprint camera image can be enhanced, thereby improving the robustness of the algorithm in various shooting scenarios.

[0093] Steps S210 to S240 are described in detail below.

[0094] In an exemplary embodiment of the present specification, the following steps can be used to reconstruct the fingerprint ridges in the contactless fingerprint photographed image through the fingerprint ridge reconstruction network to obtain the reconstructed fingerprint ridge features:

[0095] The original fingerprint ridge features in the contactless fingerprint camera image can be extracted, and the original fingerprint ridge features can be input into the fingerprint ridge reconstruction network to obtain the reconstructed fingerprint ridge features.

[0096] The original fingerprint ridge feature refers to feature data directly extracted from the contactless fingerprint photographed image and capable of characterizing the fingerprint ridge structure feature. For example, the original fingerprint ridge feature may be a high-frequency component feature extracted from the contactless fingerprint photographed image through a high-pass filter, or an image gradient feature corresponding to the contactless fingerprint photographed image, or a fingerprint ridge direction field extracted from the contactless fingerprint photographed image through an image feature extraction operator such as a Sobel operator. This embodiment is not limited thereto.

[0097] In an optional implementation, the following steps may be used to extract the original fingerprint ridge features in the contactless fingerprint photographic image, which may specifically include:

[0098] The high-frequency component data in the contactless fingerprint photographed image can be extracted by a high-pass filter based on a Gaussian kernel, and the high-frequency component data can be used as the original fingerprint ridge features corresponding to the contactless fingerprint photographed image.

[0099] Among them, high-frequency component data refers to the frequency domain representation of the fast-changing part in the contactless fingerprint camera image. The high-frequency component data in the contactless fingerprint camera image can focus on reflecting the detailed structure of the ridges and valleys of the fingerprint, while the low-frequency part in the contactless fingerprint camera image mainly reflects the background information in the contactless fingerprint camera image. By extracting the high-frequency component data in the contactless fingerprint camera image as the original fingerprint ridge feature, the expressiveness of the original fingerprint ridge feature in the contactless fingerprint camera image can be effectively enhanced, while suppressing the interference of low-frequency background noise.

[0100] The high-frequency component data in the contactless fingerprint camera image can be extracted by a high-pass filter based on the Gaussian kernel function. The high-frequency component of the contactless fingerprint camera image taken in various scenes can be extracted by controlling the Gaussian kernel. Compared with the convolution operation based on the spatial domain or the frequency domain transformation (such as discrete Fourier transform, fast wavelet transform, etc.), the accuracy and effectiveness of the high-frequency component data are improved.

[0101] Specifically, a high-pass filter constructed using a Gaussian kernel function can be used to filter the frequency components of the contactless fingerprint camera image, suppressing the low-frequency part and retaining the high-frequency part. The parameter design of the filter can include the size and standard deviation of the Gaussian kernel, and these parameters can be dynamically adjusted according to the resolution, quality, and complexity of the texture details of the contactless fingerprint camera image. For example, a larger kernel size and a smaller standard deviation are used in high-resolution images to capture more detailed information; the opposite is true for low-resolution images. For example, a high-pass filter based on a Gaussian kernel function can be expressed by equation (1):

[0102]

[0103] in, It can represent the output image after high-pass filtering, and represents the high-frequency component data obtained after removing the low-frequency component of the contactless fingerprint camera image. x can represent all the information of the contactless fingerprint camera image. The output result of x*g(r,σ) can represent the low-frequency component in the contactless fingerprint camera image. r can represent the distance from the center of the Gaussian kernel. σ can represent the standard deviation of the Gaussian kernel, which is used to control the diffusion degree of the high-pass filter.

[0104] By extracting high-frequency component data of contactless fingerprint photography images through a high-pass filter based on a Gaussian kernel, the fingerprint pattern details and background information in the contactless fingerprint photography images can be effectively separated, providing high-precision support for the local feature expression of the fingerprint image. The high-frequency component data, as the original fingerprint ridge feature, can more clearly show the local details of the ridges and valleys, while significantly reducing the interference of low-frequency background information on the feature extraction process, improving the expression accuracy of the original fingerprint ridge features, and thus ensuring the enhanced contactless fingerprint enhanced image.

[0105] In an exemplary embodiment of the present specification, the fingerprint ridge reconstruction network may include a feature encoder and a frequency domain decoder, which may be Figure 3 The steps in the above are used to input the original fingerprint ridge features into the fingerprint ridge reconstruction network to obtain the reconstructed fingerprint ridge features. Figure 3 As shown, it may specifically include:

[0106] Step S310, inputting the original fingerprint ridge features into the feature encoder, and extracting ridge structure representation data corresponding to the contactless fingerprint photographed image;

[0107] Step S320: input the ridge structure characterization data into the frequency domain decoder, and decode to obtain the reconstructed fingerprint ridge features.

[0108] Among them, the feature encoder is one of the core modules of the fingerprint ridge enhancement model. The fingerprint ridge reconstruction network and the fingerprint image reconstruction network share the same feature encoder. The task of the feature encoder is to extract multi-level ridge structure representation data from the original fingerprint ridge features. These ridge structure representation data not only contain the geometric morphology information of the fingerprint ridges, but also reflect the directional distribution and local coherence of the fingerprint lines. The feature encoder is usually composed of multiple convolutional layers, each of which extracts features of different scales and depths to achieve a comprehensive representation of the fingerprint line information.

[0109] The frequency domain decoder is a network that converts the ridge structure representation data output by the feature encoder into a restoration network of fingerprint ridge details. The frequency domain decoder can gradually restore the spatial distribution information of the fingerprint ridges through deconvolution operations or interpolation processing, and at the same time compensate for the ridge details that may be lost during the restoration and reconstruction process through residual connections.

[0110] Through the fingerprint ridge reconstruction network with the collaborative work of feature encoder and frequency domain decoder, the broken, blurred and missing ridges in the fingerprint image are effectively reconstructed and restored. The feature encoder extracts the multi-level ridge structure representation data of the contactless fingerprint camera image, providing high-dimensional feature expression for the detailed restoration of the fingerprint lines; the frequency domain decoder reconstructs the geometric form of the fingerprint lines based on residual connection to ensure the accuracy of the reconstruction and restoration of the original fingerprint ridge features. The collaborative work of the feature encoder and the frequency domain decoder enables the input original fingerprint ridge features to not only maintain the directional consistency of the local fingerprint lines, but also restore the broken ridge details through the hierarchical processing of multi-scale features. In low-contrast, blurred and noisy images, this method can stably restore the key ridge structure, improve the accuracy and stability of the reconstructed fingerprint ridge features, and provide reliable support for enhancing the overall detailed expression of the image.

[0111] In an exemplary embodiment of the present specification, the fingerprint image reconstruction network may include a feature encoder and an image decoder shared with the fingerprint ridge reconstruction network; Figure 4 The steps in the above code are used to reconstruct the contactless fingerprint camera image through the fingerprint image reconstruction network to obtain the contactless fingerprint reconstruction image. Figure 4 As shown, it may specifically include:

[0112] Step S410, inputting the original fingerprint ridge features extracted from the contactless fingerprint photographed image into the feature encoder to extract ridge structure representation data;

[0113] Step S420: input the ridge structure representation data into the image decoder to reconstruct a contactless fingerprint reconstructed image corresponding to the contactless fingerprint photographed image.

[0114] Among them, the fingerprint image reconstruction network is mainly used to reconstruct and optimize the overall information in the contactless fingerprint camera image to improve the visual clarity and recognizability of the contactless fingerprint camera image. In this network, the feature encoder extracts the ridge structure representation data that represents the texture structure by processing the input original fingerprint ridge features. These ridge structure representation data not only reflect the distribution law of local fingerprint ridges, but also combine the global illumination characteristics and background texture characteristics of the contactless fingerprint camera image, thereby providing high-quality feature input for subsequent image reconstruction.

[0115] The image decoder receives the output data of the feature encoder and generates a reconstructed contactless fingerprint reconstruction image by gradually restoring it. In the implementation process, the image decoder can generally use a multi-level deconvolution layer to spatially expand the ridge structure representation data extracted by the feature encoder, and introduce the original features in combination with jump connections to maintain the global structural consistency of the contactless fingerprint camera image. In terms of parameter setting, the number of channels and step size of the image decoder can be dynamically adjusted according to the size and resolution of the input contactless fingerprint camera image to ensure that the texture and background characteristics of the contactless fingerprint reconstruction image are clear, and the size of the reconstructed and restored contactless fingerprint reconstruction image is consistent with the size of the original image.

[0116] The global information of the contactless fingerprint camera image is reconstructed and restored through the fingerprint image reconstruction network, which can effectively improve the overall visual quality of the contactless fingerprint reconstruction image. The feature encoder extracts the global representation data of the contactless fingerprint camera image to eliminate the contrast reduction problem caused by uneven lighting or complex background, and combines with the image decoder to enhance the resolution and clarity of the contactless fingerprint reconstruction image through step-by-step restoration operations. This reconstruction method restores the separation boundary between the fingerprint pattern area and the non-fingerprint pattern area through multi-layer feature extraction and deconvolution processing, ensuring the balance between high-frequency enhancement and low-frequency smoothing of the fingerprint image. In addition, the application of residual connection and multi-scale fusion in the fingerprint image reconstruction network further improves the adaptability of the fingerprint ridge enhancement model to degraded fingerprint images, improves the accuracy and clarity of the output contactless fingerprint reconstruction image, ensures the quality of the contactless fingerprint reconstruction image, and provides higher quality image input for the subsequent fingerprint recognition process.

[0117] In an exemplary embodiment of the present specification, the following steps may be performed to achieve the fusion of the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image to obtain the contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image, which may specifically include:

[0118] The reconstructed fingerprint ridge features and the contactless fingerprint reconstruction image can be weighted fused point by point using a dynamic weighting method to obtain a contactless fingerprint enhanced image.

[0119] Among them, the dynamic weighting method is a method of weighting based on the similarity of the local features of the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image. For example, the local features of the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image can be compared, the similarity score can be calculated, and the dynamic fusion weight can be generated according to the feature similarity score, and then the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image can be weighted fused by the dynamic fusion weight.

[0120] For example, the similarity score between the reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image can be calculated using equation (2):

[0121]

[0122] Among them, S(i,j) can represent the feature similarity score between the reconstructed fingerprint ridge feature and the contactless fingerprint reconstructed image at the position (i,j), and F ridge (i, j, k) can represent the feature value of the reconstructed fingerprint ridge feature at position (i, j) and the kth channel, F image (i, j, k) may represent the feature value of the contactless fingerprint reconstruction image at position (i, j) and the kth channel, and N may represent the number of feature channels.

[0123] The dynamic fusion weight can be generated by using the relation group (3):

[0124]

[0125] W image (i,j)=1-W ridge (i,j); (3)

[0126] Among them, W ridge (i, j) can represent the dynamic fusion weight of reconstructing fingerprint ridge features, W image (i, j) can represent the dynamic fusion weight of the contactless fingerprint reconstruction image, S(p, q) can represent the similarity score at any position in the local feature comparison window, and (p, q) can represent any position in the local feature comparison window.

[0127] The reconstructed fingerprint ridge features are dynamically weighted and fused with the contactless fingerprint reconstructed image to achieve a high degree of unification of local texture features and overall image information. The dynamic weighting method based on similarity calculation can adaptively adjust the fusion weight of each pixel, making the fusion result more adaptable under complex acquisition conditions. And through the dynamic weighted fusion operation, the detailed performance of the fingerprint texture and the global consistency of the image can be optimized at the same time, effectively avoiding the fusion error caused by feature deviation in the fixed weight method. In addition, the dynamic weighting method combines the contrast information of local features, so that high-quality texture features occupy a larger weight in the enhanced image, significantly improving the recognizability and feature integrity of the contactless fingerprint enhanced image, and improving the quality of the contactless fingerprint enhanced image.

[0128] In an exemplary embodiment of this specification, it is possible to Figure 5 The steps in the pre-training of the fingerprint ridge enhancement model are implemented, refer to Figure 5 As shown, it may specifically include:

[0129] Step S510, acquiring a contactless fingerprint sample image, and performing random degradation on the contactless fingerprint sample image to obtain a contactless fingerprint simulation image;

[0130] Step S520: training the pre-built fingerprint ridge enhancement model based on the contactless fingerprint sample image and the contactless fingerprint simulation image to obtain a trained fingerprint ridge enhancement model.

[0131] The contactless fingerprint sample images refer to the original fingerprint data screened from the high-quality contactless fingerprint images actually collected, which serve as the basis for model training. By randomly degrading the contactless fingerprint sample images, the image quality degradation caused by various complex conditions in real scenes is simulated, providing a wide distribution of training samples for the model to improve the robustness and generalization ability of the fingerprint ridge enhancement model.

[0132] Random degradation refers to the generation of images based on high-quality contactless fingerprint sample images that simulate the image quality degradation caused by various complex conditions in real scenes by introducing noise and adjusting parameters to simulate different shooting scenes.

[0133] Optionally, the following steps may be used to implement random degradation of the contactless fingerprint sample image to obtain a contactless fingerprint simulation image, which may specifically include:

[0134] A preset degradation factor library can be obtained, and multiple random degradation factors can be determined based on the degradation factor library. The random degradation factors can include a degradation method and a random degradation degree corresponding to the degradation method. The contactless fingerprint sample image is degraded by the random degradation factors to obtain a contactless fingerprint simulation image that simulates the contactless fingerprint taken in a real scene.

[0135] The degradation factor library refers to a pre-set basic tool for simulating the degradation of captured images in real scenes, which may include multiple degradation modes and corresponding parameter ranges. For example, the degradation modes may include over-darkness degradation, over-exposure degradation, dirtiness degradation, motion blur degradation, out-of-focus degradation, etc. Specifically, Gaussian blur can be used to simulate the out-of-focus problem, random noise injection can be used to restore the signal noise influence of the acquisition device, and brightness and contrast adjustment can be used to simulate low-quality images captured under complex lighting conditions.

[0136] The random degradation factor is determined randomly. Each degradation process randomly selects one or more degradation modes and implements parameterization based on the degree of degradation. For example, the kernel size and standard deviation of Gaussian blur can be randomly selected within a set range to simulate out-of-focus problems in different situations. The intensity of noise injection can be generated through Poisson distribution or uniform distribution.

[0137] Figure 6 A schematic diagram of implementing degradation processing of contactless fingerprint photography images through a degradation factor library provided in an embodiment of this specification.

[0138] refer to Figure 6 As shown, for a high-quality contactless fingerprint sample image 601, the contactless fingerprint sample image 601 can be subjected to random degradation processing through a degradation factor library 602 to obtain a contactless fingerprint simulation image 603 that simulates a contactless fingerprint shot in a real scene. Specifically, the degradation processing process can be expressed by equation (4):

[0139]

[0140] Where x can represent the contactless fingerprint sample image 601, δ v (x) may represent a degradation factor library for performing degradation processing on the contactless fingerprint sample image 601, v may represent a random degradation factor selected from the degradation factor library, and V may represent the number of degradation factors.

[0141] By setting a variety of random degradation factors through the preset degradation factor library, contactless fingerprint simulation images can be generated to achieve accurate and generalized simulation of complex acquisition environments. The random combination of different degradation factors (such as blur, noise, and contrast adjustment) ensures the diversity of contactless fingerprint simulation images and provides a comprehensive distribution of degradation features for the model. This data enhancement method can guide the fingerprint ridge enhancement model to learn the feature recovery rules in a variety of complex situations, thereby significantly improving its ability to reconstruct low-quality fingerprint images. At the same time, the adaptive regulation mechanism of the random degradation factor enables the simulated image to be close to the actual acquisition conditions, providing higher quality and larger data reliability sample data for the training of the fingerprint ridge enhancement model.

[0142] Optionally, the pre-built fingerprint ridge enhancement model may include a preliminarily trained fingerprint ridge reconstruction network and a fingerprint image reconstruction network to be trained; the model training of the fingerprint ridge enhancement model may be implemented by the following steps, which may include:

[0143] The original fingerprint ridge features corresponding to the contactless fingerprint sample image can be extracted, and the simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image can be extracted; the simulated fingerprint ridge features are input into the preliminarily trained fingerprint ridge reconstruction network to obtain the reconstructed simulated ridge features, and a ridge consistency loss function is constructed based on the reconstructed simulated ridge features and the original fingerprint ridge features; the simulated fingerprint ridge features are input into the fingerprint image reconstruction network to be trained to obtain the reconstructed simulated fingerprint image, and an image consistency loss function is constructed based on the reconstructed simulated fingerprint image and the contactless fingerprint sample image; a cycle consistency loss function is constructed based on the reconstructed simulated ridge features and the reconstructed simulated fingerprint image; the overall training loss function of the fingerprint ridge enhancement model is constructed through the ridge consistency loss function, the image consistency loss function and the cycle consistency loss function, and the pre-constructed fingerprint ridge enhancement model is trained based on the overall training loss function to obtain a trained fingerprint ridge enhancement model.

[0144] Among them, the pre-built fingerprint ridge enhancement model can include a fingerprint ridge reconstruction network that has been optimized through self-supervised training and a fingerprint image reconstruction network to be further trained. The core of the structural design of the fingerprint ridge enhancement model is to use a collaborative learning strategy to make the ridge features and the overall image information complement each other during the training process.

[0145] The first step in the training process is to extract the original fingerprint ridge features and the simulated fingerprint ridge features. The original fingerprint ridge features are directly extracted from the non-degraded high-quality fingerprint samples, while the simulated fingerprint ridge features are extracted from the degraded simulated images. The original fingerprint ridge features and the simulated fingerprint ridge features are used as reference data for ridge consistency loss and image consistency loss, respectively.

[0146] The ridge consistency loss function guides model training by measuring the difference between the reconstructed simulated ridge features and the original fingerprint ridge features. In specific implementations, the loss function can be defined using pixel-level differences (such as L1 loss) or indicators based on structural similarity (such as SSIM loss) to ensure the accuracy and authenticity of ridge reconstruction.

[0147] The image consistency loss function is mainly used to measure the similarity between the reconstructed simulated fingerprint image and the original sample image. Its implementation is usually based on global comparison, combining multiple dimensions such as illumination balance and contrast consistency to evaluate image quality. The cycle consistency loss function further constrains the feature consistency of the model after multiple processing by simulating the closed-loop relationship between the input and the reconstructed output, thus avoiding feature loss or overfitting.

[0148] The overall training loss function can be achieved by weighted combination of ridge consistency loss function, image consistency loss function and cycle consistency loss function, combining ridge features, overall image information and cycle feature consistency to form a multi-objective optimization strategy. The setting of loss weights needs to be adjusted through experiments to balance the contribution of each sub-goal to training.

[0149] Figure 7 A schematic diagram of the principle of enhancing a contactless fingerprint photographic image by using a fingerprint ridge enhancement model provided in an embodiment of this specification.

[0150] refer to Figure 7 As shown, the original fingerprint ridge features 702 corresponding to the contactless fingerprint sample image 701 and the simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image 703 can be extracted; the simulated fingerprint ridge features are input into the preliminarily trained fingerprint ridge reconstruction network included in the fingerprint ridge enhancement model 704, specifically, the reconstructed simulated ridge features 707 are obtained by reconstructing and restoring the fingerprint ridge reconstruction network composed of the feature encoder 705 and the frequency domain decoder 706, and a ridge consistency loss function is constructed based on the reconstructed simulated ridge features 707 and the original fingerprint ridge features 702. For example, the ridge consistency loss function can be expressed as equation (5):

[0151]

[0152] in, can represent the ridge consistency loss function, E can represent the network parameters of the feature encoder, and D R The network parameters of the frequency domain decoder can be expressed as It can represent the original fingerprint ridge feature 702, can represent the reconstructed simulated ridge feature 707 obtained by reconstruction and restoration, ‖‖1 can represent the L1 norm, which represents the relative difference between the reconstructed simulated ridge feature 707 and the original fingerprint ridge feature 702, and is used to measure the error between the two; It can represent the expected value, which is used to represent the average error on all training samples.

[0153] Then, the simulated fingerprint ridge features can be input into the fingerprint image reconstruction network to be trained contained in the fingerprint ridge enhancement model 704. Specifically, the fingerprint image reconstruction network composed of the feature encoder 705 and the image decoder 708 is used to reconstruct and restore the reconstructed simulated fingerprint image 709, and an image consistency loss function is constructed based on the reconstructed simulated fingerprint image 709 and the contactless fingerprint sample image 701; for example, the image consistency loss function can be expressed as equation (6):

[0154]

[0155] in, can represent the image consistency loss function, x can represent the contactless fingerprint sample image 701, The reconstructed simulated fingerprint image 709 may be represented.

[0156] Then, a cycle consistency loss function can be constructed based on the reconstructed simulated ridge features 707 and the reconstructed simulated fingerprint image 709; for example, the cycle consistency loss function can be expressed as equation (7):

[0157]

[0158] in, It can be expressed as the cycle consistency loss function, It can represent the high-frequency component data obtained by performing ridge feature extraction on the reconstructed simulated fingerprint image 709 through a Gaussian filter.

[0159] Finally, the overall training loss function of the fingerprint ridge enhancement model can be constructed through the ridge consistency loss function, the image consistency loss function and the cycle consistency loss function. For example, the overall training loss function can be expressed as equation (8):

[0160]

[0161] in, The overall training loss function can be represented, and the pre-built fingerprint ridge enhancement model is trained based on the overall training loss function to obtain a trained fingerprint ridge enhancement model.

[0162] By constructing an overall training loss function that includes ridge consistency, image consistency, and cycle consistency loss functions, multi-objective optimization of the fingerprint ridge enhancement model is achieved. The ridge consistency loss ensures the consistency between the geometric shape of the reconstructed ridge and the original ridge features; the image consistency loss improves the global quality of the reconstructed image; and the cycle consistency loss further constrains the closed-loop relationship between feature input and output to avoid feature loss. This training method, through the joint optimization of multiple loss functions, can not only improve the fingerprint ridge enhancement model's ability to restore the details of ridge features, but also enhance its adaptability to global image quality, so that the fingerprint ridge enhancement model can show high-precision feature recovery effects in multiple degradation scenarios, providing reliable support for subsequent fingerprint recognition or identity authentication.

[0163] In an exemplary embodiment of this specification, it is possible to Figure 8 The steps in the above code implement the self-supervised training of the fingerprint ridge reconstruction network. Figure 8 As shown, it may specifically include:

[0164] Step S810, extracting original fingerprint ridge features corresponding to the contactless fingerprint sample image, and extracting simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image;

[0165] Step S820, using the original fingerprint ridge features as self-supervisory signals and the simulated fingerprint ridge features as inputs, the pre-constructed fingerprint ridge reconstruction network is subjected to self-supervisory training to obtain a preliminarily trained fingerprint ridge reconstruction network.

[0166] Among them, self-supervised training refers to a training method that does not require manually labeled data. In this embodiment, the self-supervised training of the fingerprint ridge reconstruction network uses the original fingerprint ridge features as high-quality self-supervised signals to guide the fingerprint ridge reconstruction network to reconstruct and restore low-quality simulated fingerprint ridge features into high-quality original fingerprint ridge features.

[0167] By using the original fingerprint ridge features as self-supervisory signals, the pre-built fingerprint ridge reconstruction network is trained in a self-supervisory manner, so that the fingerprint ridge reconstruction network can be trained without manually annotated data. The introduction of self-supervisory signals enhances the ability of the fingerprint ridge reconstruction network to reconstruct and restore the fingerprint ridge structure details, while retaining the global structural consistency of the fingerprint pattern. The self-supervised training process can significantly reduce the dependence on high-quality annotated data, allowing the fingerprint ridge reconstruction network to learn the mapping relationship between degraded fingerprint images and high-quality images in more diverse unlabeled data, thereby improving the ability to restore low-quality simulated fingerprint ridge features.

[0168] Fig. 9 A schematic diagram of the principle of self-supervised learning of a fingerprint ridge reconstruction network provided in an embodiment of this specification.

[0169] refer to Fig. 9 As shown, the original fingerprint ridge features 902 corresponding to the contactless fingerprint sample image 901 and the simulated fingerprint ridge features 903 corresponding to the contactless fingerprint simulation image can be extracted, and then the original fingerprint ridge features 902 are used as self-supervisory signals and the simulated fingerprint ridge features 903 are used as input to perform self-supervisory training on the pre-constructed fingerprint ridge reconstruction network. Specifically, the simulated fingerprint ridge features 903 are input into the feature encoder 904 of the fingerprint ridge reconstruction network to extract ridge structure representation data, and the ridge structure representation data is input into the frequency domain decoder 905, and the reconstructed fingerprint ridge features 906 are obtained by decoding. The original fingerprint ridge features 902 are used as self-supervisory signals, and the difference between the original fingerprint ridge features 902 and the reconstructed fingerprint ridge features 906 is minimized through the ridge consistency loss function 907. The network parameters of the pre-constructed fingerprint ridge reconstruction network are optimized and adjusted to obtain a preliminarily trained fingerprint ridge reconstruction network.

[0170] In summary, the contactless fingerprint image enhancement method provided in the embodiments of the present specification inputs the contactless fingerprint photographed image into a pre-trained fingerprint ridge enhancement model, the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network, and reconstructs the fingerprint ridges in the contactless fingerprint photographed image through the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features; reconstructs the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image; and fuses the reconstructed fingerprint ridge features into the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image. On the one hand, the fingerprint ridges in the contactless fingerprint camera image are reconstructed through the fingerprint ridge reconstruction network included in the fingerprint ridge enhancement model, and the contactless fingerprint camera image is enhanced by combining the reconstructed fingerprint ridge features obtained by reconstruction, which can improve the quality of key fingerprint features in the contactless fingerprint camera image in a targeted manner, thereby improving the accuracy of the contactless fingerprint enhanced image; on the other hand, the reconstructed fingerprint ridge features obtained based on the reconstruction are fused with the contactless fingerprint reconstructed image to obtain the final contactless fingerprint enhanced image, which can effectively improve the quality and clarity of the key fingerprint features in the contactless fingerprint enhanced image, namely the fingerprint ridges. Even if the quality of the input contactless fingerprint camera image is poor, the key features in the contactless fingerprint camera image can be enhanced, thereby improving the robustness of the algorithm in various shooting scenarios; on the other hand, the fingerprint ridge enhancement model can realize the enhanced processing of the contactless fingerprint camera image through a single round of reasoning, with a small amount of calculation, low latency, and saving of computing resources. It can be run on mobile devices with low computing power, and can meet the real-time requirements and portability requirements of identity authentication product reasoning.

[0171] See also Fig.10, which is a flow chart of a contactless fingerprint identity authentication method according to an embodiment of this specification. In the embodiment of this specification, the contactless fingerprint identity authentication method can be applied to a terminal device or a server, and this exemplary embodiment does not make any special restrictions on this. Fig.10 The process shown in the figure is described in detail. The contactless fingerprint identity authentication method in the embodiment of this specification may specifically include the following steps:

[0172] Step S1010, acquiring a contactless fingerprint photographed image, and inputting the contactless fingerprint photographed image into a pre-trained fingerprint ridge enhancement model to obtain a contactless fingerprint enhanced image;

[0173] Step S1020: performing a comparison and matching in a preset identity identification database based on the contactless fingerprint enhanced image to obtain an identity authentication result of the contactless fingerprint photographed image.

[0174] The fingerprint ridge enhancement model can be used to reconstruct and restore contactless fingerprint photographed images, and obtain high-quality and high-definition contactless fingerprint enhanced images. The contactless fingerprint enhanced images contain clear and easily recognizable fingerprint ridge features, which can realize accurate recognition of contactless fingerprint enhanced images and use them for identity authentication, thereby improving the accuracy of identity authentication results.

[0175] See also Fig.11 , is a schematic diagram of the structure of a contactless fingerprint image enhancement device provided in an embodiment of this specification. Fig.11 As shown, the contactless fingerprint image enhancement device 1100 can be implemented as all or part of an electronic device through software, hardware or a combination of both. According to some embodiments, the contactless fingerprint image enhancement device 1100 includes a fingerprint image input module 1110, a fingerprint ridge reconstruction module 1120, a fingerprint image reconstruction module 1130, and a modal feature fusion module 1140, specifically including:

[0176] The fingerprint image input module 1110 is used to obtain a contactless fingerprint photographic image and input the contactless fingerprint photographic image into a pre-trained fingerprint ridge enhancement model, wherein the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network;

[0177] The fingerprint ridge reconstruction module 1120 is used to reconstruct the fingerprint ridges in the contactless fingerprint photographed image through the fingerprint ridge reconstruction network to obtain the reconstructed fingerprint ridge features;

[0178] The fingerprint image reconstruction module 1130 is used to reconstruct the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image;

[0179] The modal feature fusion module 1140 is used to fuse the reconstructed fingerprint ridge features into the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image.

[0180] Optionally, the fingerprint ridge reconstruction module 1120 is configured to:

[0181] Extracting original fingerprint ridge features from the contactless fingerprint photographed image;

[0182] The original fingerprint ridge features are input into the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features.

[0183] Optionally, the fingerprint ridge reconstruction module 1120 is configured to:

[0184] Extracting high-frequency component data from the contactless fingerprint photographed image by using a high-pass filter based on a Gaussian kernel;

[0185] The high frequency component data is used as the original fingerprint ridge features corresponding to the contactless fingerprint photographed image.

[0186] Optionally, the fingerprint ridge reconstruction network includes a feature encoder and a frequency domain decoder, and the fingerprint ridge reconstruction module 1120 is configured as follows:

[0187] Inputting the original fingerprint ridge features into the feature encoder to extract ridge structure representation data corresponding to the contactless fingerprint photographed image;

[0188] The ridge structure characterization data is input into the frequency domain decoder and decoded to obtain the reconstructed fingerprint ridge features.

[0189] Optionally, the fingerprint image reconstruction network includes a feature encoder and an image decoder shared with the fingerprint ridge reconstruction network; the fingerprint image reconstruction module 1130 is configured as follows:

[0190] Inputting the original fingerprint ridge features extracted from the contactless fingerprint photographed image into the feature encoder to extract ridge structure representation data;

[0191] The ridge structure characterization data is input into the image decoder to reconstruct a contactless fingerprint reconstruction image corresponding to the contactless fingerprint photographed image.

[0192] Optionally, the modality feature fusion module 1140 is configured as follows:

[0193] The reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image are weightedly fused point by point using a dynamic weighting method to obtain a contactless fingerprint enhanced image;

[0194] The dynamic weighting method is a method of allocating weights based on the similarity between the reconstructed fingerprint ridge features and the local features of the contactless fingerprint reconstructed image.

[0195] Optionally, the contactless fingerprint image enhancement device 1100 further includes a fingerprint ridge enhancement model training module, and the fingerprint ridge enhancement model training module is configured as follows:

[0196] Acquire a contactless fingerprint sample image, and randomly degrade the contactless fingerprint sample image to obtain a contactless fingerprint simulation image;

[0197] The pre-constructed fingerprint ridge enhancement model is trained based on the contactless fingerprint sample image and the contactless fingerprint simulation image to obtain a trained fingerprint ridge enhancement model.

[0198] Optionally, the fingerprint ridge enhancement model training module is configured as:

[0199] Acquire a preset degradation factor library, and determine a plurality of random degradation factors based on the degradation factor library, wherein the random degradation factors include a degradation mode and a random degradation degree corresponding to the degradation mode;

[0200] The contactless fingerprint sample image is degraded by the random degradation factor to obtain a contactless fingerprint simulation image that simulates the contactless fingerprint captured in a real scene.

[0201] Optionally, the pre-built fingerprint ridge enhancement model includes a preliminarily trained fingerprint ridge reconstruction network and a fingerprint image reconstruction network to be trained; the fingerprint ridge enhancement model training module is configured as follows:

[0202] Extracting original fingerprint ridge features corresponding to the contactless fingerprint sample image, and extracting simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image;

[0203] Inputting the simulated fingerprint ridge features into a preliminarily trained fingerprint ridge reconstruction network to obtain reconstructed simulated ridge features, and constructing a ridge consistency loss function based on the reconstructed simulated ridge features and the original fingerprint ridge features;

[0204] Inputting the simulated fingerprint ridge features into a fingerprint image reconstruction network to be trained to obtain a reconstructed simulated fingerprint image, and constructing an image consistency loss function based on the reconstructed simulated fingerprint image and the contactless fingerprint sample image;

[0205] Constructing a cycle consistency loss function based on the reconstructed simulated ridge features and the reconstructed simulated fingerprint image;

[0206] The overall training loss function of the fingerprint ridge enhancement model is constructed by the ridge consistency loss function, the image consistency loss function and the cycle consistency loss function, and the pre-constructed fingerprint ridge enhancement model is trained based on the overall training loss function to obtain a trained fingerprint ridge enhancement model.

[0207] Optionally, the fingerprint ridge enhancement model training module is configured as:

[0208] Extracting original fingerprint ridge features corresponding to the contactless fingerprint sample image, and extracting simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image;

[0209] The original fingerprint ridge features are used as self-supervisory signals and the simulated fingerprint ridge features are used as inputs to perform self-supervisory training on the pre-constructed fingerprint ridge reconstruction network to obtain a preliminarily trained fingerprint ridge reconstruction network.

[0210] See also Fig.12 , is a schematic diagram of the structure of a contactless fingerprint identity authentication device provided in an embodiment of this specification. Fig.12 As shown, the contactless fingerprint identity authentication device 1200 can be implemented as all or part of an electronic device through software, hardware, or a combination of both. According to some embodiments, the contactless fingerprint identity authentication device 1200 includes a fingerprint image enhancement module 1210 and a fingerprint image verification module 1220, specifically including:

[0211] The fingerprint image enhancement module 1210 is used to obtain a contactless fingerprint photographic image, and input the contactless fingerprint photographic image into a pre-trained fingerprint ridge enhancement model to obtain a contactless fingerprint enhanced image;

[0212] The fingerprint image verification module 1220 is used to compare and match the contactless fingerprint enhanced image in a preset identity identification database to obtain an identity verification result of the contactless fingerprint photographed image.

[0213] The above device embodiments correspond to the method embodiments. For specific descriptions, please refer to the description of the method embodiments, which will not be repeated here. The device embodiments are obtained based on the corresponding method embodiments and have the same technical effects as the corresponding method embodiments. For specific descriptions, please refer to the corresponding method embodiments.

[0214] The embodiments of this specification also provide a computer storage medium, which can store multiple instructions, and the instructions are suitable for being loaded by a processor and executing the method as described in the embodiments of this specification. The specific execution process can be found in the specific description of the embodiments of this specification, and will not be repeated here.

[0215] This specification also provides a computer program product, which stores at least one instruction, and the at least one instruction is loaded by the processor and executes the method as described in the embodiments of this specification. The specific execution process can be found in the specific description of the embodiments of this specification, and will not be repeated here.

[0216] The embodiments of this specification also provide Fig.13 The structural diagram of the electronic device shown in FIG. Fig.13 At the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to implement the above-mentioned voice activity detection method.

[0217] Of course, in addition to software implementation, this specification does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is to say, the executor of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0218] For the improvement of a technology, it can be clearly distinguished whether it is a hardware improvement (for example, improvement of the circuit structure of diodes, transistors, switches, etc.) or a software improvement (improvement of the method flow). However, with the development of technology, many improvements of the method flow today can be regarded as direct improvements of the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented with a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming themselves, without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0219] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.

[0220] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0221] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0222] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0223] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0224] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0225] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0226] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0227] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0228] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0229] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0230] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0231] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0232] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0233] The above description is only an embodiment of the present specification and is not intended to limit the present specification. For those skilled in the art, the present specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification shall be included in the scope of the claims of the present specification.

Claims

1. A contactless fingerprint image enhancement method, the method comprising: Acquire a contactless fingerprint photographic image, and input the contactless fingerprint photographic image into a pre-trained fingerprint ridge enhancement model, wherein the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network; Reconstructing the fingerprint ridges in the contactless fingerprint photographed image by the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features; Reconstructing the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image; The reconstructed fingerprint ridge features are fused to the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image.

2. According to the contactless fingerprint image enhancement method of claim 1, the step of reconstructing the fingerprint ridges in the contactless fingerprint photographed image by the fingerprint ridge reconstruction network to obtain the reconstructed fingerprint ridge features comprises: Extracting original fingerprint ridge features from the contactless fingerprint photographed image; The original fingerprint ridge features are input into the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features.

3. According to the contactless fingerprint image enhancement method of claim 2, the step of extracting original fingerprint ridge features from the contactless fingerprint photographed image comprises: Extracting high-frequency component data from the contactless fingerprint photographed image by using a high-pass filter based on a Gaussian kernel; The high frequency component data is used as the original fingerprint ridge features corresponding to the contactless fingerprint photographed image.

4. According to the contactless fingerprint image enhancement method of claim 2, the fingerprint ridge reconstruction network includes a feature encoder and a frequency domain decoder, and the inputting the original fingerprint ridge features into the fingerprint ridge reconstruction network to obtain the reconstructed fingerprint ridge features comprises: Inputting the original fingerprint ridge features into the feature encoder to extract ridge structure representation data corresponding to the contactless fingerprint photographed image; The ridge structure characterization data is input into the frequency domain decoder and decoded to obtain the reconstructed fingerprint ridge features.

5. According to the contactless fingerprint image enhancement method of claim 1, the fingerprint image reconstruction network comprises a feature encoder and an image decoder shared with the fingerprint ridge reconstruction network; the step of reconstructing the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain the contactless fingerprint reconstructed image comprises: Inputting the original fingerprint ridge features extracted from the contactless fingerprint photographed image into the feature encoder to extract ridge structure representation data; The ridge structure characterization data is input into the image decoder to reconstruct a contactless fingerprint reconstruction image corresponding to the contactless fingerprint photographed image.

6. The contactless fingerprint image enhancement method according to claim 1, wherein the step of fusing the reconstructed fingerprint ridge features into the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image comprises: The reconstructed fingerprint ridge features and the contactless fingerprint reconstructed image are weightedly fused point by point using a dynamic weighting method to obtain a contactless fingerprint enhanced image; The dynamic weighting method is a method of allocating weights based on the similarity between the reconstructed fingerprint ridge features and the local features of the contactless fingerprint reconstructed image.

7. The contactless fingerprint image enhancement method according to claim 1, further comprising: Acquire a contactless fingerprint sample image, and randomly degrade the contactless fingerprint sample image to obtain a contactless fingerprint simulation image; The pre-constructed fingerprint ridge enhancement model is trained based on the contactless fingerprint sample image and the contactless fingerprint simulation image to obtain a trained fingerprint ridge enhancement model.

8. The contactless fingerprint image enhancement method according to claim 7, wherein the step of randomly degrading the contactless fingerprint sample image to obtain a contactless fingerprint simulation image comprises: Acquire a preset degradation factor library, and determine a plurality of random degradation factors based on the degradation factor library, wherein the random degradation factors include a degradation mode and a random degradation degree corresponding to the degradation mode; The contactless fingerprint sample image is degraded by the random degradation factor to obtain a contactless fingerprint simulation image that simulates the contactless fingerprint captured in a real scene.

9. According to the contactless fingerprint image enhancement method of claim 7, the pre-constructed fingerprint ridge enhancement model comprises a preliminarily trained fingerprint ridge reconstruction network and a fingerprint image reconstruction network to be trained; The method of training the pre-built fingerprint ridge enhancement model based on the contactless fingerprint sample image and the contactless fingerprint simulation image to obtain a trained fingerprint ridge enhancement model includes: Extracting original fingerprint ridge features corresponding to the contactless fingerprint sample image, and extracting simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image; Inputting the simulated fingerprint ridge features into a preliminarily trained fingerprint ridge reconstruction network to obtain reconstructed simulated ridge features, and constructing a ridge consistency loss function based on the reconstructed simulated ridge features and the original fingerprint ridge features; Inputting the simulated fingerprint ridge features into a fingerprint image reconstruction network to be trained to obtain a reconstructed simulated fingerprint image, and constructing an image consistency loss function based on the reconstructed simulated fingerprint image and the contactless fingerprint sample image; Constructing a cycle consistency loss function based on the reconstructed simulated ridge features and the reconstructed simulated fingerprint image; The overall training loss function of the fingerprint ridge enhancement model is constructed by the ridge consistency loss function, the image consistency loss function and the cycle consistency loss function, and the pre-constructed fingerprint ridge enhancement model is trained based on the overall training loss function to obtain a trained fingerprint ridge enhancement model.

10. The contactless fingerprint image enhancement method according to claim 9, further comprising: Extracting original fingerprint ridge features corresponding to the contactless fingerprint sample image, and extracting simulated fingerprint ridge features corresponding to the contactless fingerprint simulation image; The original fingerprint ridge features are used as self-supervisory signals and the simulated fingerprint ridge features are used as inputs to perform self-supervisory training on the pre-constructed fingerprint ridge reconstruction network to obtain a preliminarily trained fingerprint ridge reconstruction network.

11. A contactless fingerprint identity authentication method, the method comprising: Acquire a contactless fingerprint photographed image, and input the contactless fingerprint photographed image into a pre-trained fingerprint ridge enhancement model to obtain a contactless fingerprint enhanced image; Based on the comparison and matching of the contactless fingerprint enhanced image in a preset identity identification database, an identity authentication result of the contactless fingerprint photographed image is obtained.

12. A contactless fingerprint image enhancement device, comprising: A fingerprint image input module, used to obtain a contactless fingerprint photographic image, and input the contactless fingerprint photographic image into a pre-trained fingerprint ridge enhancement model, wherein the fingerprint ridge enhancement model includes a fingerprint ridge reconstruction network and a fingerprint image reconstruction network; A fingerprint ridge reconstruction module, used to reconstruct the fingerprint ridges in the contactless fingerprint photographed image through the fingerprint ridge reconstruction network to obtain reconstructed fingerprint ridge features; A fingerprint image reconstruction module, used to reconstruct the contactless fingerprint photographed image through the fingerprint image reconstruction network to obtain a contactless fingerprint reconstructed image; The modal feature fusion module is used to fuse the reconstructed fingerprint ridge features into the contactless fingerprint reconstructed image to obtain a contactless fingerprint enhanced image corresponding to the contactless fingerprint photographed image.

13. A contactless fingerprint identity authentication device, the device comprising: A fingerprint image enhancement module is used to obtain a contactless fingerprint photographic image and input the contactless fingerprint photographic image into a pre-trained fingerprint ridge enhancement model to obtain a contactless fingerprint enhanced image; The fingerprint image verification module is used to compare and match the contactless fingerprint enhanced image in a preset identity identification database to obtain the identity verification result of the contactless fingerprint photographed image.

14. A 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 10 or 11 are implemented.

15. An electronic device, characterized in that: include: A processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the steps of the method as claimed in any one of claims 1 to 10 or 11.

16. A computer program product having at least one instruction stored thereon, characterized in that: When the at least one instruction is executed by the processor, the steps of the method described in any one of claims 1 to 10 or 11 are implemented.

Images