Research On Individual Recognition Based On Fingerprint And Palm Print Biometric Fusio

Following the development of artificial intelligence,contact fingerprint,and palmprint biometric individual recognition has formed a certain scale both in the field of scientific research and in commercial applications.However,with the development of human society and the continuous iterative challenges(such as health security and other issues),contactless fingerprint and palmprint feature recognition technology is more novel than traditional contact fingerprint and palmprint recognition technology and has more advantages in some fields.In addition,the performance of fingerprint or palmprint recognition in single mode is not as good as that of multi-mode fusion recognition in processing sample nonlinear deformation,anti-counterfeiting recognition,and even final recognition accuracy.Based on the above,the author carried out this research.The general work content and contribution of this paper are as follows:1)In terms of contactless fingerprint recognition,an RMO(Ridge and Minutiae Optimization)algorithm based on ridge structure is proposed in this paper.The algorithm eliminates the pseudo-feature points based on repairing the pseudo-ridge.Secondly,contactless fingerprint recognition lacks adaptive and efficient feature extraction and matching algorithms.In this paper,a BCS(Base Circle Structure)model and an IPM(Incentive & Punishment Mechanism)matching algorithms are proposed.The BCS model constructs the base circle structure through the feature points and their neighborhood and is used to extract the eigenvector of the feature points in the ROI(Region of Interest).IPM compensates the matching score and obtains the final matching result by using the feature point information twice.In an experiment based on a self-built contactless fingerprint dataset,this paper compares similar methods and discusses the results.Based on the experimental data,it is proved that the RMO algorithm proposed in this paper is effective in fingerprint feature extraction and the combination of IPM and BCS model can greatly improve the accuracy of fingerprint recognition.In the comparison experiment of recognition algorithms,the EER(Equal Error Rate)of this scheme is equal to 0,which is significantly better than the excellent similar methods in recent years.2)In terms of Contactless fingerprint recognition,this paper first proposes a set of highly adaptable RSC(ROI Segmentation of Contactless-palmprint)algorithms in the process of contactless palmprint sample pretreatment.Based on certain prior knowledge,this algorithm extracts palmprint ROI through the traditional methods of key point positioning and proportional segmentation.Secondly,to improve the recognition rate of contactless palmprints and avoid the problem of long time delay caused by traditional classical recognition algorithms,this paper studies the characteristic of palmprints,and in the search and description stage of palmprint feature points,An Effective Contactless-palmprint Feature(ECF)based on local feature description is proposed.This method includes a lightweight palmprint feature point search algorithm and two palmprint feature description methods based on neighborhood intensity difference and neighborhood gradient respectively.In an experiment based on a self-built contactless palmprint dataset,a comparison experiment of similar methods is conducted and the results are discussed and analyzed.Among them,the effective area of contactless palmprint ROI extraction based on the RSC algorithm accounted for 96.58%.At the same time,compared with classical target recognition algorithms and excellent similar algorithms in recent years,the overall performance of contactless palmprint recognition based on the ECF algorithm is better than that of the comparison methods listed in the EER=0.35%.3)In terms of contactless fingerprint and palmprint fusion recognition,based on the current research in this field,two IQA(Image Quality Assessment algorithms)were designed based on SURF feature point search and ROI frequency domain intensity respectively.In principle,the above methods are used to evaluate the quality of target ROI at both local and global levels.According to this evaluation mechanism,the quality score of samples is calculated and weight is allocated.Finally,score-fusion is carried out for fingerprint and palmprint recognition.Based on the experiment of two public contactless data sets of fingerprint and palmprint,the data are discussed and analyzed.The ROI quality evaluation algorithm of fingerprints and palmprints proposed in this paper can effectively improve the RR(Recognition Rate)of fingerprints or palmprints under the premise of limited time loss.Secondly,in the experimental comparison with similar advanced methods,the contactless fingerprint and palmprint fusion recognition based on quality evaluation in this paper shows excellent performance of RR=100 with a delay of 195.5ms.