A Fast and Efficient Palmprint Identification Method for a Large Database

Abstract

This paper proposes a fast and efficient palmprint identification method for a large database. The process is accelerated as a result of our efficient palmprint classification and matching scheme. Palmprint classification method is based on principle lines which are life line, head line and heart line. Palmprints’ features are extracted with Log-Gabor filter and matched with Hamming distance in the most potential palmprint group/category, and if necessary, continues orderly to the less potential ones. Experiments are done with 2 hand databases, Visgraph database and CU-CGCI hand database. Experimental results show that the proposed method can greatly reduce the number of template matching from 100% (as in general identification methods) to 33.2-38.2% while maintaining the equivalent EER as the general identification method.