An intelligent authentication system using wavelet fusion of K-PCA, R-LDA

In this work, we proposed a novel authentication system based on facial features. The proposed method is based on PCA and LDA for feature extraction, these extracted features are combined using wavelet fusion. In this work we use neural networks to classify extracted features of faces. The proposed method consists of six steps: i) Extraction of images from the database, ii) Preprocessing, iii) Feature extraction using PCA, iv) feature extraction using LDA, v) Wavelet fusion of the extracted features, extracted from PCA and LDA and, vi) classification using neural network. Features are extracted using both PCA and LDA to improve capability of LDA when few samples of images are available. Wavelet fusion and neural networks are used to improve classification accuracy. The proposed system shows improvement over the existing methods particularly when the database contains occluded images. Preliminary experimental results have shown high accuracy of the system.