Study And Realization On Mobile Terminals Oriented Partial Fingerprint Matching Algorithm

At present,most mobile terminals use small fingerprint sensors,which can only capture part of fingerprint images,and the inputted partial fingerprint images cannot provide sufficient minutiae features.Therefore,the traditional minutia-based fingerprint matching algorithm cannot be applied to mobile terminals.In addition,mobile applications such as mobile payment using fingerprint identification systems put higher requirements on fingerprint matching algorithms.It is imperative to study a new,more accurate,more secure and more efficient fingerprint matching algorithm for mobile terminals.This paper is inspired by the conventional fingerprint matching algorithm based on Minutia Cylinder-Code(MCC)structure.Through analysis and experiment,this paper proposes a partial fingerprint matching algorithm based on Modified Feature Point Cylinder Code(MFPCC)structure.The MFPCC makes two improvements to the original MCC.First,the modified structure halves the layers number of the original MCC by changing the contribution calculation method of the original MCC structure.While ensuring the matching performance,the computational complexity is greatly reduced,and the original MCC structure redundancy and construction time are solved.Second,the modified structure extends the feature selection range of the original MCC.It can be combined with minutiae to form the Modified Feature Point Cylinder Code-Minutia(MFPCC-M),and it can also be combined with ridge points to form the Modified Feature Point Cylinder Code-Ridge Point(MFPCC-RP).The problem of feature loss caused by too small sensor area is solved.The partial fingerprint matching algorithm based on MFPCC is divided into three stages:feature structure establishment,local matching and global matching.In the feature structure establishment stage,MFPCC-M and MFPCC-RP are established respectively for minutiae and ridge points;in the local matching stage,the local similarity is calculated using the binary vector correlation for the MFPCC-M structure and the MFPCC-RP structure respectively;in the global matching stage,the Local Similarity Assignment with Relaxation(LSA-R)algorithm is used to obtain the global score.The algorithm maintains the characteristics of the original MCC.It has invariant characteristics for translation and rotation,and robustness to distortion and slight spurious feature points.Besides,it is binary coded and.low reversible.In addition,it meets the requirements of partial fingerprint identification systems on mobile terminals for tolerance of fingerprint image direction,robustness of fingerprint skin state,matching efficiency and fingerprint template security.The proposed algorithm is tested on the FVC2002 and FVC2004 databases and compared with four well-known matching algorithms.The results show the proposed method has excellent comprehensive performance in equal error rate,security and matching speed,and has outstanding ability to apply to light architecture.