An Approach to Face Recognition Using Feed Forward Neural Network

Abstract

Many approaches have been proposed for face recognition but there are major constraints like illumination, lightning, pose etc., when taken into consideration, results in poor recognition rate. We propose a method to improve the recognition rate of the face recognition system which uses various methods like homogeneity, energy, covariance, contrast, asymmetry, correlation, mean, standard deviation, entropy, kurtosis to extract the facial features for a better recognition rate. Also the extracted features are trained and it is associated with a feed forward back propagation neural network used for classification to render better results.