Suitability study of certain features and classifiers for Bharatanatyam double-hand mudra images

Bharatanatyam, a traditional Indian classical dance, requires a study under the supervision of experts. At present, there is a scarcity of experts in this fine art, which requires leveraging of technology to make it self-pursuable. This paper presents a three-stage methodology for classification of double-hand mudra images of Bharatanatyam, wherein contours of mudras are obtained first, cell features are extracted that include number of vertical and horizontal intersections of grid lines with the contours of the mudras, mudra type and Hu-moments, second and classified using three classifiers, namely, rule based, artificial neural network and k-nearest neighbour, third. A comparative study of classifiers is given for their suitability. The proposed method finds many applications such as e-learning of mudras and proper postures leading to self-learning of Bharatanatyam dance, and online commentary during concerts.