Research On Fingerprint Recognition Matching Algorithm

Automatic fingerprint verification technology became the earliest and most widely used biometric authentication technology due to the uniqueness and invariability of fingerprint. In the past two decades, automatic fingerprint verification technology has developed rapidly and has been applied in various fields. In this paper, based on the research on many literature of automatic fingerprint identification system,three aspects-------minutiae post-processing, matching algorithms and registrationsystem are studied.1. Introduce the history of fingerprint verification system development as well as its work flow.2. Due to the fact that many false minutiae exist after the process of extraction, we discuss on a post-processing algorithm to delete false minutiae using neighborhood information. The experiment results show that this algorithm can delete most of the false minutiae and improve the recognition performance.3. Using ridge tracing algorithm to delete false minutiae and store true minutiae at the same time can improve the efficiency of the algorithm.4. In the second stage of matching, we used changeable bounding boxes to fulfill the fuzzy matching of minutiae, which can avoid the position and angle deviation of the minutiae caused by local deformation of the fingerprint image.5. Implement registration system by merging features of three template fingerprint. The algorithm is tested on fingerprint register machine based on Samsung ARM7 processor S3C44BOX. False rejection rate is 22% and false accept rate is 0.18%.