Nonlinear Fingerprint Orientation Smoothing by Median Filter

Abstract

The fingerprint local ridge orientation is a kind of very important intrinsic property in fingerprint image processing and recognition. Getting a reliable orientation field is still a great challenge in poor quality fingerprint caused by scars and ridge breaks. This paper employs a median filter to attenuate impulsive-like noise in the orientation field such as the scars and ridge breaks, etc. In addition, a reliability measure of estimated orientation by the median filter is proposed and an adaptive orientation smoothing algorithm is developed to compute a reliable orientation field based on varying smoothing windows. The experimental results of some synthetic images and real poor quality fingerprints demonstrate that our proposed approach works well to attenuate noise especially the impulsive-like noise gracefully while maintaining the orientation localization of the high-curvature area