Performance Evaluation of Contourlet Transform based Palmprint Recognition using Nearest Neighbour Classifier
K. Shanmugapriya, Karthika M S, S. Valarmathy, Munikrishna Arunkumar · 2013
Abstract- Biometrics-based personal verification is a powerful security features in technology era. Palmprint recognition is an accepted and widely used biometric. Palmprint is an important complement and reliable biometric that can be used for identity verification because it is stable and unique for every individual. This paper presents a new palmprint matching method by using the contourlet features. The region of interest is extracted from the palmprint image as a preprocessing step. The contourlet transform is a new two dimensional extension of the wavelet transform using multi-scale and directional filter banks. It can effectively capture smooth contours, that is both the global and local details which are the dominant features in palmprint images. The large number of coefficients generated from the contourlet transform are minimised as a dimensionality reduction process, by calculating energies for each subband. Energy features are calculated as a feature selection step and feature vector is created. Palmprint matching is then performed using nearest neighbor classifier. Each subband is compared with the corresponding subband by simple nearest neighbour classifier. Among subbands of each image, the majority voting scheme is followed for classification.The proposed method improves classification accuracy and reduces computation time.