Abstract
This paper presents theoretically simple, yet computationally efficient approach for face recognition. There are many transforms and entropy measures used in face recognition technology. Recognition rate is poor with binary and edge based recognition techniques. We employ the entropy concept to binary and edge images. We use Conventional Entropy Quantifier (CEQ) which counts only the transitions, and Modified Entropy Quantifier (MEQ) which considers the positions with transitions for measuring the entropy. The proposed entropy features possess good texture discriminative property. The experiments are conducted on benchmark databases using SVM and K-NN classifiers. Experimental results show the effectiveness of our system.