A simple and novel fingerprint image segmentation algorithm

Fingerprint image segmentation is initial part of fingerprint recognition system. Fingerprint segmentation separates fingerprint area from image background. Segmentation is used to avoid extraction of feature in noisy areas that are common in background. In this paper we propose a very simple algorithm that uses Laplacian of Gaussian (LoG) filter, edge filter (LoG based) and morphological operations. This algorithm does not require block wise computation of mean and variance. Proposed algorithm also take advantage of the fact that most fingerprint images are vertical or within 45 degrees to the vertical. The proposed algorithm is applied on FVC2004-DB1, DB2, DB3 and FVC2000-DB3 public database. Experimental results show that algorithm gives good output even for low quality images.