Fingerprint Image Features Extraction Using Moment Invariants

Abstract

The extraction of features of fingerprint image is the most important in fingerprint identification and verification. The robustness of the system is based totally upon its reliably extracted feature from images. Representing the large amounts of data that lay in a fingerprint image requires an efficient method of feature extraction. More fingerprint image acquisition is vulnerable to unpredicted errors born of deforming operations like rotation, translation of the fingerprint and scaling. Image representation is, therefore, required to be invariant with the deformations and maintain the discriminating features of the individual. Moments are to be considered a very qualifying descriptor should the tradeoff between the discriminating power and the invariant be taken into account. This paper proposes a new algorithm for the feature extractions; globally and locally from fingerprint images depending on moment invariants.