Discrete Hidden Markov Model based Face Recognition System Introducing Sustainability of Uneven Lighting Distortion

Abstract

The aim of this work is to enhance the performance of appearance based face recognition system to remove the uneven lighting effect which is generally occurred in natural environment. Active Shape Model has been used to detect the facial shape and contrast based edge detection technique has been applied to reduce the uneven lighting problem. Linear Discriminant Analysis has been used to reduce the dimension of the facial feature vector. Discrete Hidden Markov Model has used for learning and testing model for create template for each human face and recognize those pattern. Extended Yale face database has used to measure the performance of the proposed face recognition system with the existence of various uneven lighting effects. Experimental results and performance analysis show superiority of the proposed system with different variations of light with the rate of change of azimuth and elevation.

Keywords: Face Recognition, Uneven Lighting Effect, Appearance based feature, Linear Discriminant Analysis, Discrete Hidden Markov Model