Face Recognition Using Edge Information and DCT

Abstract

This manuscript presents a face identification method based on the theory of Sobel Local Binary Pattern and Laplacian filters edge detectors to represent face images with enriched information. Thereafter, to refine images for less sample size, Discrete Cosine Transform is used to represent these face images in low dimension space. The proposed method of facial recognition performs well to a considerable amount under small sample problem and is also invariant to illumination changes. It shows better results as compared to previously developed face recognition methods after performing experiments on many well known databases.