SME feature-based classification of Adavu and posture images of Bharatanatyam dance

The Adavus are typical and fundamental postures in Bharatanatyam dance. Adavu is a predefined sequence of hands, legs and neck with pre-set body pose forms, which are studied under the guidance of experts, who are becoming scarce in these days. This paper proposes a three-stage approach for classification of Adavu images. In the first stage, contours of Adavus are obtained. In the second stage, the structural(S), moments(M) and eigen values(E) features, namely, intersections, symmetric property, width_height_difference, contour distance, Hu-moments and eigenvalues are obtained. ANN is used for classification of Adavus in third stage. A comparative study of classification accuracies of using different features is made. The envisaged applications of this research include e-learning of Adavus and Bharatanatyam dance, in particular, and various dances, in general, evaluation of Adavus exhibited by a performer for their accuracies, and automation of commentary during the concerts and the like.