Fuzzy Shannon Entropy for Face Recognition

This paper presents a new local feature called Fuzzy Shannon entropybased features for face recognition. These features are computationally simple and capture the variation in the face image. The new Fuzzy membership functions are defined which are not like standard Gaussian or trapezoidal membership functions, but they are computed with normalized Image intensity values. This paper gives the flexibility to design of new membership functions to cater the need of the problem. The Information set is the combination of fuzzy membership and Information source (attributes). By applying this concept, we have extracted features from ORL and Faces94 databases with Support Vector Machine and K Nearest Neighbour classifier the results obtained with the proposed method are better than other published contemporary techniques.