Face Recognition Based on PCA and LDA Combination Feature Extraction

Abstract

A whole face recognition system was proposed in the paper based on PCA and LDA combination feature extraction. Normalization was used to eliminate the redundant information interference. Principal Component Analysis (PCA) was used for feature extraction and dimension reduction. Linear Discriminate Analysis (LDA) was used to further improve the separability of samples in the subspace and extract LDA features. Nearest Neighbor Classifier (NNC) was adopted for face recognition. Face recognition rate was improved in our experiments on ORL face database with our approach.