Face recognition against variation in pose and background

Human faces are full of information but their recognition by computer is a complex task. Recognition of faces is a routine task for humans and it is normally done by using various face features such as eyes, color of eyes, skin color, positions and shapes of nose, mouth and shapes of hair and its color. Keeping track of all this information is difficult for a computer system. Hence, only certain important and unique features are isolated and used for recognition of face. Here, we have attempted to produce an efficient principal component analysis and artificial neural network system to identify faces. The system gives 100% results for standard database for variation in pose and background. The system is also tested for locally created poor image quality database and its performance is satisfactory.