A biometric based personal Identification system using face

Abstract

This thesis proposes models for automatic face recognition using newlinelocal region and component based approaches The main focus of the work is newlineto understand global and local feature extraction techniques for face newlinerecognition system to achieve good recognition rate newline Three types of face recognition models are proposed in the work newlineThe first approach is based on modular localized variation using Eigen space newlineApproach also called Localized Principal Component Analysis LPCA for newlineface recognition This algorithm is compared with traditional PCA and newlineModular PCA MPCA for face images with large variations in illumination newlinepose expression and partial occlusions In the proposed approach dividing newlinethe images into adequately smaller modules will help in localizing the facial newlinevariations so that the classification ability can be improved The localization newlineof those variations gets better with smaller modules If the modules become newlinesmaller and smaller then the dependencies among pixels may be ignored To newlineovercome these problems the modules should have some standard form by newlineapplying module creation strategy The accuracy of the LPCA produces better newlineresults in the classification phase newlineThe second type of approach adopts local component approach newlinecalled Random Image Component RIC to overcome large facial variations newline newline newline