Fingerprint Reference Point Detection

Abstract

A robust fingerprint recognition algorithm should tolerate the translation and rotation of fingerprint image. This paper proposes an algorithm to locate a unique reference point and compute a unique reference orientation for all types of fingerprints, which can be used for translational and rotational alignments. The reference point localization is based on scale space analysis of Poincare index (PI) and orientation variance (OV), and the reference orientation is computed based on analysis of orientation differences between the radial directions from the reference point and the local ridge orientations along these radials. In addition, a nonlinear smoothing method with better performance in filtering impulse noise and the noise caused by scars and ridge breaks etc. is proposed to compute the orientation fields of varying scales except for the finest scale. Experimental results demonstrate that the proposed algorithm can locate the reference point and compute the reference orientation with high accuracy for all types of fingerprints.