Research On High-Resolution Fingerprint Image Quality Enhancement Method Based On Residual Network

To a certain extent,the quality of fingerprint image determines the accuracy of fingerprint feature extraction and feature recognition,which has a great impact on the performance of fingerprint recognition system.Therefore,it is very important to design appropriate and effective fingerprint quality enhancement methods.There are a lot of pores in high-resolution fingerprint images,and fingerprint recognition based on pores can achieve higher safety performance.Poor-quality fingerprint images will face problems such as unstable pores extraction results and insufficient number of pores matching under different images of the same fingerprint.However,the existing fingerprint quality enhancement methods mainly focus on denoising and improving the clarity of fingerprint ridge and valley line,and do not pay enough attention to the information of pores.In summary,this paper discusses and studies the quality enhancement of high-resolution fingerprint image.In this paper,a high-resolution fingerprint image quality enhancement method based on residual network is proposed.The residual network is introduced into the fingerprint image quality enhancement problem.The residual branch and the low-frequency information branch are combined.The residual branch is used to extract and integrate the deep high-frequency features of the fingerprint image,and the low-frequency information branch is used to extract and retain the low-frequency information of the input fingerprint image.The high-frequency information of the high-resolution fingerprint is combined with the low-frequency information by enlarging and adding the two branches at the same time,which effectively avoids the generation of pseudo pores and pseudo details,and improves the ability of reconstructing fingerprint details.For the residual blocks in the residual branch,a wider pattern is adopted to avoid the loss of information and promote the better delivery of information flow.A lot of experimental results show that the method can not only improve the image quality effectively,but also improve the stability of pore extraction and the accuracy of recognition.At the same time,this method also greatly reduces the time consumption in fingerprint enhancement stage.This paper further studies the high-resolution fingerprint quality enhancement method based on residual network and prior features of fingerprints.The purpose of this method is to introduce the influence of prior features of fingerprint into the enhancement method,improve the performance of the model through the guidance and constraint of prior features,and reconstruct more and clearer fingerprint details.In this method,residual in residual is used to avoid the difficulty of retaining low frequency of fingerprint and solve the model degradation problem with the deepening of the network.The global information and local information are more fully coupled by the combination of long connection and short connection.The attention module is introduced into the residual in residual structure to enhance the sensitivity of the network to each channel and different spatial position of fingerprint image,and strengthen the ability of feature filtering.