High-resolution Fingerprint Image Pore Detection Method

With the development of Times,information security is getting more and more attention,which has raised higher requirements about the related equipment.Due to the upgrade of hardware and software technology,the acquisition of the third layer fingerprint characteristic information is possible.Based on high-resolution fingerprint image's characteristics of pores,the fingerprint identification and authentication system has strong anti-counterfeiting and living detection ability and is security,popular in mobile device miniaturization.The traditional high-resolution fingerprint image detection algorithm is mainly based on skeleton refinement and matching filter method,affected by the high-resolution fingerprint image quality or filtering model,the extraction effect of fingerprint pore is not desirable.Based on the high-resolution fingerprint image data collected in the real environment,this paper improves the dynamic anisotropic fingerprint pore model algorithm to adapt to different high-resolution fingerprint image data.The extraction result of fingerprint pore is better than that of the high-resolution fingerprint image data used in the paper of dynamic anisotropic pore model.In this paper,a high-resolution fingerprint image pore extraction algorithm based on SURF features is proposed.The SVM classification model was used for training.Experimental result show that is better than that of the improved algorithm based on dynamic anisotropic pore model.Deep learning is more and more widely used in image processing.The high-resolution fingerprint image pore is detected by using the Faster RCNN detection network.The experimental results verify the feasibility of deep learning method in high-resolution fingerprint image pore extraction.