Enhanced Minutiae Extraction for High-Resolution Palmprint Recognition

Abstract

A palmprint generally possesses about 10 times more minutiae features than a fingerprint, which could provide reliable biometric-based personal authentication. However, wide distribution of various creases in a palmprint creates a number of spurious minutiae. Precisely and efficiently, minutiae extraction is one of the most critical and challenging work for high-resolution palmprint recognition. In this paper, we propose a novel minutiae extraction and matching method for high-resolution palmprint images. The main contributions of this work include the following. First, a circle-boundary consistency is proposed to update the local ridge orientation of some abnormal points. Second, a lengthened Gabor filter is designed to better recover the discontinuous ridges corrupted by wide creases. Third, the principal ridge orientation of palmprint image is calculated to establish an angle alignment system, and coarse-to-fine shifting is performed to obtain the optimal coordinate translation parameters. Following these steps, minutiae matching can be efficiently performed. Experiment results conducted on the public high-resolution palmprint database validate the effectiveness of the proposed method.