Research On Low Quality Fingerprint Restoration Based On Partial Differential Equation

With the rapid development of social information security,biological information has been widely concerned,and fingerprint has become the most widely used biological information because of its uniqueness,universality,permanence,reliability and other characteristics.However,in practical application,especially in some criminal cases,the collection of fingerprint information is affected by the equipment and environment,and the human fingers are wet and loose Due to skin,oil,static and other reasons,part of the extracted fingerprint images are often low-quality fingerprint images with fuzzy,broken ridge and complex background.This kind of low-quality fingerprint image will reduce the recognition rate of the fingerprint identification system,and even lead to unjust and false cases in some criminal cases.Most of the low-quality fingerprint images contain more pseudo feature points and potential feature points,but the existing low-quality fingerprint image processing methods have poor effect on the restoration of potential feature points,so the restoration of low-quality fingerprint image has become a hot topic The solution of this problem will not only improve the accuracy of fingerprint identification equipment,but also have great significance for the judicial fingerprint comparison.Aiming at the characteristics of low quality fingerprint information and the large number of breakpoints,the traditional low quality fingerprint image processing method has been improved.The improved fingerprint model is added to the low quality fingerprint image based on the traditional preprocessing algorithm,and the fingerprint image is advanced and denoised and repaired.It reduces the interference caused by complex background to the repair algorithm.The accuracy of repair is improved.The main achievements and innovations are as follows:1.According to the research purpose of this paper,research and optimize the traditional fingerprint preprocessing algorithm;according to the characteristics of lowquality fingerprint,use the segmentation algorithm,Gabor enhancement algorithm and binarization algorithm in the traditional fingerprint preprocessing algorithm to process the low-quality fingerprint image,and eliminate the thinning algorithm in the traditional preprocessing algorithm,in order to reduce the background noise of lowquality fingerprint image and reduce the repair cost The difficulty of the algorithm,improve the accuracy of the repair.2.Learn the traditional partial differential equation model,and reproduce the three classical partial differential equation repair models: bsbc repair model,TV repair model and CDD repair model.After comparison,CDD repair model is selected as the basic algorithm of this subject,and an exponential factor and adaptive positive lifting parameter are introduced into the classical CDD repair.After experimental comparison,the improved CDD repair algorithm is not only effective It inherits the advantages of the classical CDD algorithm and reduces the difficulty of the algorithm.3.According to the shortcomings of traditional partial differential equation(PDE)in repairing fingerprint texture,the improved CDD repair model and the MCA repair model proposed by Starck et al are fused and improved.After verification,the final algorithm in this paper not only solves the false phenomenon of the classical PDE image repair model in repairing texture region,but also overcomes the lack of original MCA repair model The method satisfies the disadvantage of connectivity.4.The realization of fingerprint image processing software mainly uses vs2013,qt5.9 and opencv3.0 as development tools,and applies the improved fingerprint preprocessing algorithm and fingerprint image restoration algorithm to the system,realizing the rapid restoration of low-quality fingerprint image.