Advanced Partial Palmprint Matching Based on Repeated Adjoining Minutiae

Abstract

Nowadays, high resolution palmprint images are used for recognition. The features that can be extracted from a high resolution palmprint image include the minutiae points. In this paper, instead of full palmprints, partial palmprints are used for matching. Partial refers to a part of the palmprint such as the thenar and hypothenar or hypothenar and interdigital areas. The minutiae can be easily located from the thinned palmprint image by using a window. Since there are a large number of minutiae present within a palmprint image, the minutiae are grouped into several clusters. The extracted minutiae are clustered using Hough circles. In order to avoid spurious minutiae resulting from the presence of immutable creases, radon transform is made use of. By selecting initial minutiae pairs, the entire matching is done by using repeated adjoining minutiae matching. The algorithm is developed and successfully tested with palmprint database.