Performance Evaluation Of Selected Principal Component Analysis-Based Techniques For Face Image Recognition

Abstract Principal Component Analysis PCA is an eigen-based technique popularly employed in redundancy removal and feature extraction for face image recognition. In this study performance evaluation of three selected PCA-based techniques was conducted for face recognition. Principal Component Analysis Binary Principal Component Analysis BPCA and Principal Component Analysis Artificial Neural Network PCA-ANN were selected for performance evaluation. A database of 400 50x50 pixels images consisting of 100 different individuals each individual having 4 images with different facial expressions was created. Three hundred images were used for training while 100 images were used for testing the three face recognition systems. The systems were subjected to three selected eigenvectors 75 150 and 300 to determine the effect of the size of eigenvectors on the recognition rate of the systems. The performances of the techniques were evaluated based on recognition rate and total recognition time.The performance evaluation of the three PCA-based systems showed that PCA ANN technique gave the best recognition rate of 94 with a trade-off in recognition time. Also the recognition rates of PCA and B-PCA increased with decreasing number of eigenvectors but PCA-ANN recognition rate was negligible.