IMPROVED FACE RECOGNITION TECHNIQUES

Abstract

Face recognition presents a challenging problem in the field of image analysis and computer vision. In the presented work we have proposed a novel approach for face recognition using Scale-invariant feature transform.We have the standard ORL(Olivetti Research Laboratories) Database. It contains20 subject having 10 different orientation and expression. The database is divided into two parts, training and testing database. The SIFT features are generated for every training image and the features called key points is calculated, then k nearest neighbour’s classifier is used for the matching scheme for test data . The recognition results demonstrate its robust performance under different expression conditions, Pose variation, illumination changes and partial occlusion. The Equal error rate which can be calculated by FAR and FRR is 0.80 and the recognition rate is 97.91% that show the robustness of the proposed method.