New enhancement algorithm for fingerprint images

Abstract

In this paper, an improved algorithm for enhancement of fingerprint image is proposed on the basis of the image normalization and Gabor filter Firstly, the adaptive normalization based on block processing is suggested for improvement of fingerprint images. An input image is partitioned into sub-blocks with the size of K /spl times/ L at first and the region of interest (ROl) of the fingerprint image is acquired The parameters for the image normalization are adaptively determined according to the statistics of each block. By utilizing these parameters, the block image is normalized for the next process. Secondly, a new technique for selection of two important parameters of Gabor filter is devised. These parameters are the ridge direction and the ridge frequency. In this study, the ridge direction of a block image is determined by the probabilistic approach unlike other works. With this ridge direction, the ridge frequency is selected by utilizing the directional projection. The proposed algorithms are tested with NIST fingerprint images and show significant improvement in the experiments.