A Comparative Study of Suitability of Certain Features in Classification of Bharatanatyam Mudra Images Using Artificial Neural Network

Abstract

Bharatanatyam is an Indian classical dance, which is composed of various body postures and hand gestures. This ancient art of dance has to be studied under the supervision of experts but at present there is dearth of Bharatanatyam dance experts. This has led to take leverage of technology to make this dance self pursuable. Thus, it is the motivation for automation of identification of mudras through image processing. This paper presents a 3-stage methodology for classification of single hand mudra images. The first stage involves acquisition and preprocessing of images of mudras to obtain contours of mudras using canny edge detector. In the second stage, the features, namely, Hu-moments, eigenvalues and intersections are extracted. In the third stage artificial neural network is used for classification of mudras. The comparative study of classification accuracies of using different features is provided at the end. To corroborate the obtained classification accuracies, a deep learning approach, namely, convolutional neural network is adopted. The work finds application in e-learning of ‘Bharatanatyam’ dance in particular and dances in general and automation of commentary during concerts.