Application of neural network for face recognition
Aung Aung Phyo
Date of Issue2005
School of Electrical and Electronic Engineering
In this dissertation, we investigate the face recognition performance of Principal Component Analysis (PCA) Face Recognition method and Radial Basis Function Neural Network Face Recognition method. Also, the effects of different training numbers of images per person are also studied in our dissertation. The PCA program and RBF NN program are tested. The ORL face database was used and we split into 2, 4, 6 and 8 images per person randomly picked for the training set and the rest for test set. The PCA method has 4.63% error rate but the RBF NN classifier only has 1.25% error rate when using 50 component feature vectors. When we use 20 component feature vectors, the PCA method has 5.63% error rate but the RBF NN classifier only has 2% error rate. Experimental results indicate that the RBF NN classifier has better performance for face recognition system more than PCA at least with respect to the ORL face database.
DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Biometrics
Nanyang Technological University