Face Recognition using LDA based Support Vector Machine

Abstract

Abstract
Recognition of a person is an easy task for human brains. From different experiments, we have found that even one to three days old babies are able to distinguish between known faces. So, it is very difficult task for a computer to recognize an unknown human face although a computer has been trained with a huge set of human face. Now-a-days face recognition has received significant attention of researchers. Lots of improvements have been obtained in successful face recognition. Support Vector Machine is a method to classify things, which is implemented successfully on face recognition. Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) are basic method, widely used in feature extraction of faces. This paper presents an approach to face recognition system using Support Vector Machine with LDA based facial features. Two
large facial databases have been used to measure the performance of the system. Linear Discriminant Analysis is applied to find aspects of faces, which are important for identification and verification. It is considered that large eigen values are adequate in face recognition system. LDA is used for dimension reduction, which is done by selecting eigenfaces of nth largest eigenvalues. Finally, these eigenfaces are projected onto face space and used for training in classification using SVM classifier.

Keywords: Face recognition, support vector machine, linear discriminant analysis, facial feature vector