Personal identification using knuckleprint

Abstract

A novel biometric defined as "knuckleprint" is presented in this paper. The Line feature of the knuckleprint with its distribution in the finger (which is defined as location feature) is extracted to identify a person. To enhance the performance of identification, hierarchical classification method is used to classify the location feature and line feature in different levels. Though this is the first attempt of knuckleprint identification, the accuracy rate reaches 96.88% on the database that contains 1,432 image samples, which testifies that knuckleprint is reliable as a biometric, and demonstrates the effectiveness and robustness of the features. © Springer-Verlag Berlin Heidelberg 2004.