Beam Search Algorithm for Fingerprint Reference Point Determination Based on Joint Orientation Features

In this paper, we propose an efficient beam search algorithm for fingerprint reference point determination with its joint orientation features, which operates in two-step as well as two-path. For the two-path operation, it uses different orientation features at different scales block-based orientation certainty and pixel-based orientation map. While some fault symptoms are roughly pruned by a block-based orientation feature which is used to ignore some local maxima and select some candidate regions for a reference point region and an accurate path to a reference point is guided by a pixel-based orientation map, their information also is fused for making up one anothers weakness. For the two-step operation, in the first step, several candidates of a reference point are determined through beam search algorithm and, in the second step, an accurate reference point is located through the specic concept models. The proposed system has pixel-based precision in determining the position of reference point without complex enhancement, which has been successfully tested through the database FVC2004.