Efficient palmprint search based on database clustering for personal identification

Abstract

This paper proposes an efficient palmprint searching algorithm for personal identification based on database clustering, which reduces the search space of fine matching. A complex filter is applied to double orientation field to detect the symmetry of the palm lines as the main feature at coarse-level search. A K-means clustering technique is applied to partition the symmetry feature space into clusters. A query processing is proposed to facilitate an efficient searching. The experimental results on the public database of Hong Kong Polytechnic University show the effectiveness of the proposed searching algorithm.