A Matlab based Face Recognition GUI system Using Principal Component Analysis and Artificial Neural Network
Achala Khandelwal, Jaya Sharma · International Journal of Modern Trends in Engineering and Research · 2016
With the advances in use of face recognition techniques for the security and surveillance system, it has become a challenging task to develop a system of face recognition which is highly efficient in use. Face images are similar in overall configuration and so face recognition is difficult. Humans are very powerful in recognizing the faces that they see often. With the advent of very high performance computing, a computer can be made to mimic large amount of interconnections and networking that exist between all the nerve cells in a human brain. This can be done by implementing a network of artificial neurons which functions in the same way as of actual neural network of human brain. Of all the biometrics applications face recognition is more important as it does not require any physical interaction of the person to be recognized. Face recognition system can be used and developed in two ways: the identification or the verification. Identification corresponds to one to many recognition i.e. the unknown face is matched with the many faces given in the database and the matched one is displayed as the output. Verification corresponds to one to one interaction i.e. the unknown face is verified as whether it is face of the specified person or not. In this paper, we have dealt with the identification of the unknown faces using ORL database. Face recognition techniques can be approached in two ways. (i) Appearance based (ii) Feature based. In the appearance based techniques, the whole face image is considered and the features are extracted through it. In the feature based technique, some of the features of face (like eyes, nose, mouth etc) are being considered for feature extraction. The feature based approaches include elastic bunch graph etc. The appearance based includes PCA, LDA, ICA etc. Compared to appearance based, feature based techniques of face recognition are less sensitive to variations in illumination. The eigenface approach or the Principal component analysis is the appearance based technique.