Hand shape based biometric authentication system using radon transform and collaborative representation based classification

Biometric authentication systems are used to identify individuals based on their unique physiological and behavioral characteristics for access control and security enhancement. In this paper, we propose a novel method of authentication using hand images by collaborative representation based classification (CRC). The contour or hand shape of a query image is extracted by morphological operations and its radon transform is computed along an optimal direction to produce unique one-dimensional feature vector. The feature vector of the query image is then coded over similarly processed training samples from all subjects (or classes) and identified as a member of the class which produces the least reconstruction residual by regularized least square (RLS) instantiation of collaborative representation. Extensive experiments were conducted on a database of 300 images employing two different mathematical operators for dimensionality reduction of acquired feature vector.