Face in a real time system. It gained importance

Face recognition is one form of biometric identification which has become an important way of recognizing person in a real time system. It gained importance due to its applications in many security related areas such as at airport terminals, authentication. Recognition is a method of identifying a person in still image or in video. From literature we can come across different traditional methods. New research of recognition in this area in recent years is focussed towards using neural network methods. Moving a step ahead in research in this area is application of Fuzzy system. In this comparative study our focus was on neural networks, fuzzy system and neuro-Fuzzy system.Keywords— Neural networks, face recognition, fuzzy system, Neuro-fuzzy.1.INTRODUCTIONFace is a physical characteristic that a person posses to identify him/her self automatically. Face recognition is a process of identifying an image by matching with the images in the database. Face Detection is the process of claiming a person as an authorized user. Its a type of one – to –one matching. Facial images may be represented as a 2D or 3D. The 3D facial images are able to identify pose variations, has no effect due lightening. The main challenges one has to encounter in face recognizing are due to facial expressions, rotation effects, noise and distortion. Some times recognition is complex due to interclass and intra class similarities. In the first type two persons may appear to be similar such as twins where as in the second case same person but change in pose, facial expression aging effects appears to be of different. Face recognition can be done in four different ways. UsingKnowledge Based Approach— 1, 2, 3, 4 we encode faces based on rules. In Feature Invariant method we identify the facial features that doesn’t change due to expressions, pose, illumination etc. Hand coded templates are stored and then used for face detection in Template Matching which is the simplest of all methods. The images are trained using different learning methods of neural networks, PCA, SVM inAppearance Based Method— Most of the research in face recognition in the recent years are based on the last method. From 2 the steps in face recognition are acquiring the face image, pre-processing is done to remove any noise in the image. Later two sets of data are selected, one for training and one for testing the new Image. The performance of any face recognition system is determined by the parameters False Acceptance Rate(FAR) – number of times a unauthorized users are accepted, False Rejection Rate (FRR) – number of times an authorized user has been rejected, Equal Error rate(EER)- FAR and FRR equals, Time to verify, Time taken to capture.In this study Section 2 consists of Face recognition methods using Neural Networks, Section 3 focuses on Fuzzy System. Section 4 describes a combination of both approaches Nuero-Fuzzy for face detection. In section 5 a comparative analysis is made on the two approaches that is neural networks and fuzzy system based on parameters. Section 6 we conclude on the Neuro-Fuzzy system

BACK TO TOP
x

Hi!
I'm Al!

Would you like to get a custom essay? How about receiving a customized one?

Check it out