PCA and LDA-based face verification using back-propagation neural network
Lih-Heng Chan, Sh‐Hussain Salleh, Chee‐Ming Ting, A. K. Ariff · 2010
In this paper, we present back-propagation neural network (BPNN) as back-end classifier for face verification. Face features are extracted based on principal component analysis (PCA) and linear discriminant analysis (LDA). PCA efficiently reduces dimension of face images and represent them with eigenfaces; while LDA is alternatively used to improve discriminant ability of the PCA algorithm. Back-propagation neural network (BPNN) is used to learn the patterns of PCA and LDA features and produce relevant client and imposter scores for verification. The algorithms were evaluated using AT&T face database which comprises 40 subjects and with a total size of 400 images. Experimental results show that BPNN significantly improves the performance of face verification which is based on Euclidean distance. Percentages of improvement in equal error rate (EER) by range 62%–85% is achieved by BPNN.