AN EFFICIENT BIOGEOGRAPHY BASED FACE RECOGNITION ALGORITHM

Abstract

Extracting the optimal features from images is always required in face recognition algorithms to achieve high accuracy. In this thesis we have presented an efficient face recognition algorithm based on Biogeography Based Optimization (BBO). We use Principal Component Analysis (PCA) as the face recognition technique to extract the most desirable features from the image. This is because an image doesn’t require all the data present in it for its representation. These extracted features are minimal required features to represent an image. First we apply Gabor filters to the correct the images to be given as input to PCA. Gabor filters help in proper alignment of images. Then we extract the features present in the images with the help of PCA. Finally we apply BBO to select the most desirable features out of the features already selected by PCA. The proposed biogeography based face recognition algorithm is applied to search the most desirable features based on well defined fitness function. Performance analysis is performed using Olivetti research Laboratory (ORL) face database. Performance results show that biogeography based face recognition algorithm generates better results than original PCA technique with gabor filters.