FACE RECOGNITION: NEURAL-NETWORK APPROACHES

There is a crucial need for high security, with data and information accumulating in abundance. More attention has now been given to biometrics. Face biometrics, useful for the authentication of a person, is a simple and non-intrusive method that recognizes face in a complex multidimensional visual model and develops for it a computational model. Faces are complex, multidimensional, meaningful visual stimuli and it is difficult to develop a computational model for face recognition. We present a hybrid neural network solution that matches other methods favorably. The system combines local image sampling, a neural map network that is self-organizing, and a convolutional neural network. The self-organizing map provides a quantization of image samples into a topological space where nearby inputs in the original space are also nearby in the output space, thereby reducing the dimensionality and invariance of minor image sample changes, and the convolutionary neural network provides partial unchanged to translation, rotation, scale, and deformation. First we present an overview of face recognition in this paper and discuss the methodology and how it works. Then we represent the latest techniques of face recognition listing their advantages and disadvantages. Some techniques specified here also improve the effectiveness of face recognition underdifferent conditions of lighting and expression of face images. We use a 400 image database of 40 people that contains quite a high degree of variability in expression, pose, and facial details. We analyze the complexity of computations and discuss howto add new classes to the trained recognize