Reliable Orientation Field Estimation of Fingerprint Based on Adaptive Neighborhood Analysis

Abstract

Fingerprint Orientation estimation is an important step in feature extraction and classification. However, a reliable extraction of fingerprint orientation data is still a challenge for poor quality images. In this paper, a gradient based estimation of orientation field based on the analysis of orientation consistency in the neighborhood for regularizing the orientation field is proposed. Experimental results are analyzed and compared with other existing gradient based methods used in this work. Evaluation performed on standard FVC2002 fingerprint databases DB1, DB2 and sample fingerprint images collected using optical fingerprint reader exhibit visibly better orientation estimation for various quality images using the proposed method.