A novel scheme for fingerprint enhancement

A fingerprint image is composed of a foreground and background areas. The foreground area is composed of ridges and valleys, while the background area is the region that surrounds the fingertip and is mainly noisy. Fingerprint recognition techniques amount to extracting minutiae as well as identifying ridge and valley regions. As the captured fingerprint images may be of poor quality due to various reasons, it is necessary to enhance the image prior to minutiae extraction. Usually, fingerprint images are enhanced either in the spatial or frequency domains. However, due to the complicated ridge structures, the quality of the enhanced images is not satisfactory. In this paper, a novel enhancement method is proposed. It makes use of a recently developed fingerprint segmentation technique, in order to separate the fingerprint foreground and background areas, and subsequently speeds up the enhancement process. Next, a novel interpolation-based technique is proposed to join broken ridges. Finally, in order to estimate the ridge-valley spatial spacing, an accurate Radon-based technique is proposed. This accurate spacing is necessary for the design of 2-D directional Gabor filter that removes artifacts, and isolated black regions. The enhancement algorithm has been implemented and tested on fingerprint images drawn from FVC2002 and FVC2004 databases. A qualitative measure of the enhanced performance is also described. Simulation results show that our proposed algorithm has verified significant quality improvement over the existing published techniques.