COLORED PATTERN IMAGE CLASSIFICATION USING NEURAL NETWORKS

Abstract

Face Recognition from imagesis a sub-area of the general object recognition problem. Artificial Neural Networks (ANN) is an automatically nonstatistical approach. ANN is studied and implemented to solve face recognition problem. The objective . ofthe present work is to use the modified linear data reduction network• for image coding (feature extraction) and a hybrid neural network system for recognition. Feature extraction methods can be classified • into linear and non-linear data reduction. Principal Component analysis (PCA}is considered as modified linear neural network based• on Karhunen-Loeve projection. Also we introduce Kohonen Self-Organizing Feature Map (SOFM) neural network as a non-linear data reduction method.• The performance ofhybrid neural network system was investigated using both gray-scale and color information of the input face images. •