Perception-based fingerprint image enhancement

People can easily recognize and correct discontinuities in images of regular patterns, such as broken ridges in fingerprint images. By utilizing this idea of human perception, we present in this paper a new fingerprint image enhancement method. The proposed method includes pattern partitioning and blockwise enhancement suitable for fingerprint images. In particular, we divide the fingerprint image into blocks of variable sizes and then apply on each block our pattern partitioning, which uses 20 log-Gabor filters and nearest neighbor clustering. Experimental results show that the proposed method successfully enhances the test fingerprint images, especially from the viewpoint of connecting broken ridges.