A new entropy function for feature extraction with the refined scores as a classifier for the unconstrained ear verification

For high end security like surveillance there is a need for a robust system capable of verifying a person under the unconstrained conditions. This paper presents the ear based verification system using a new entropy function that changes not only the information gain function but also the information source values. This entropy function displays peculiar characteristics such as splitting into two modes. Two types of entropy features: Effective Gaussian Information source value and Effective Exponential Information source value functions are derived using the entropy function. To classify the entropy features we have devised refined scores (RS) method that refines the scores generated using the Euclidean distance. The experimental results vindicate the superiority of proposed method over literature.