Palm-line detection

Abstract

Palm lines, which consist of principal lines and wrinkles, are stable and essential traits for palmprint-based individual identification and can be extracted in low-resolution images. However, the research on palm-line detection has done little. Due to special properties of palmprint, in addition to the structure feature, width of the palm-line, which generally reflects strength information, is important to identify palms especially when various palmprints have similar structures. In this paper, a palm-line detection approach is proposed to simultaneously extract structure and strength features of palm lines by minimizing a local image area which is of similar brightness to each individual pixel. The presented method has been tested on the PolyU Palmprint Database and compared with the Canny edge detector and SUSAN edge finder. Experimental results illustrate the effectiveness of this approach.