期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2014
卷号：2
期号：8
出版社：S&S Publications
摘要：Impressions of fingers lifted from crime scenes (latent prints) identification is complex task and isextremely important to law enforcement agencies. Partial fingerprints are usually called latents with small area,nonlinear distortion, and are usually blurred and smudgy. Because of these characteristics and fetures, they have asignificantly smaller number of minutiae points (one of the most important features in fingerprint matching) andtherefore it can be extremely difficult to automatically match latents to plain or rolled fingerprints that are stored in lawenforcement databases. The main aim is to develop a latent matching algorithm that uses only minutiae informationunder noise. The proposed approach consists of following three modules: (i) align two sets of minutiae by using adescriptor-based Hough Transform; (ii) establish the correspondences between minutiae; and (iii) compute a similarityscore under three cases noise, noiseless and by Noise reduction
