Face recognition method using mutual information and hybrid feature

Abstract

A two-stage face recognition algorithm is proposed. In the first stage, the mutual information match is applied to reduce the candidate pattern amount in the database. In the second stage, the principal component analysis (PCA) and linear discriminant analysis (LDA) are applied to the probe image to extract the corresponding features, which will be used in classification. The approach is tested on ORL database, Shimon database and Harvard database. The experimental results demonstrate that the performance of this system is superior to other algorithms on the same database.