Face Recognition Using Neural Network and PCA



Face Recognition Using Neural Network and PCA

Introduction

Advancement in technology and a rising saleable demand have noticed an increase in the demand for a system for facial recognition, that can be useful for security reasons, that is strong by the movement of head in a plane, and also on the conditions of lightening. A technique for the same face recognition purpose is proposed that makes utilization of both Neural Networks and PCA that is Principal Component Analysis (Zhao et.al, 2003). The proposed technique includes extracting features of the face that are significant after that subjecting it to Principal Component Analysis in order to lessen the dimensions, and at the end categorizing the facial features by making use of neural networks. The algorithm's outcomes will be gathered and contrasted to any other algorithm that has been tested on the identical dataset. Techniques of decreasing the FRR or false recognition rate and FAR or false acceptance rate will be examined, and if the structure of the neural network is changed the performance will be improved (Viennet et.al, 1993). This proposal offers a step by step way to conduct the research along with a motivation and time frame of the research.

Research Question

Designing a system for face recognition utilizing Neural Networks with Modular PCA that is healthy be means of changing conditions of lighting, the face's lateral offset in the image, and variations like the variations in hair style as well as addition of spectacles.

Analyzing the rate of false acceptance as well as false rejection. With examining the techniques of decreasing these errors.

Research Hypothesis

It is possible to design a system for face recognition using Neural Networks and Modular PCA that will offers an improved rate of recognition than any other systems.

Discussion

Face recognition may be considered as one of the complicated problem in the field of computer science. The advancement in technology in the recent years has permitted us to significantly improve our infrastructure as well as algorithms for the face recognition. At the same time as there are some other booming biometric systems for identification existing now, the thing is none of them have the extensibility and freedom of the face recognition. At present face recognition is yet in its early years, and is not efficient or robust sufficient to be effectively implemented in production (Sue-Kyeong et.al, 2011). Several new moves towards the face recognition have turn out to be available in recent times, though the older methods have not worn out their prospective. While if we see face recognition in the context of security purposes, than the system would come across its own specific issues. These issues are connected to the input data, in which, we will not have or wish to have the assistance of an individual. This shows the individual may be recognized without possessing to grasp their face in a specific way, or strength an exacting facial expression (Singh et.al, 2012).

Conditions of lighting can possibly change, not only in the light, excluding also the way ...