Fuzzy L Membership Function Based Hand Gesture Recognition for Bharatanatyam Dance

This paper presents a method for automatic hand gesture recognition of ´Bharatanatyam´ dance using Fuzzy L membership function based approach. Here, a 3-stage system has been designed. In the first stage, the hand of the dancer from background is isolated using Texture based segmentation and thus the contour of the hand is extracted by using Sobel edge detection technique. In the next stage, centre point of the boundary is located and based on this eight spatial distances are calculated. These distances are normalized by dividing the maximum distance value. In the final stage, fuzzy L Membership values are calculated for each distance and matching of an unknown hand gesture is done with the known hand gestures from the database based on L fuzzy membership function. The proposed algorithm gives overall an accuracy of 85.1% and timing complexity is 2.563 sec in an Intel Pentium Dual Core processor running Mat lab R011b for each hand gesture. This simple yet effective code is very useful for e-learning of ´Bharatanatyam´ dance.