Bio-Metric Identification Utilizing Principal Component Analysis And Piece Coordinate Based Classifiers With Divided Approach

Abstract

A new approach of segmented face recognition which improves identification through better L2 Norm based
distance classifier is proposed on this paper. The enrolled images are segmented into face blocks and face recognition
algorithm is performed based on the blocks. By segmenting the image in to blocks we can identify and eliminate the problems
of local occlusion and distortion in face images. The improved distance classifier which is able to select the best matched
blocks and eliminates the affected blocks by providing appropriate weight for classifying each block depending upon the
knowledge based on images simulated in particular condition