Bharatanatyam hand gesture recognition using polygon representation

For automatic hand gesture recognition of `Bharatanatyam´ dance, a 4-stage system has been designed. In the first stage, texture based segmentation is done to detect the hand of the dancer from the background. Thus boundary of the hand is extracted. In the next stage, boundary of the hand is approximated using straight lines. Now at the third stage, each straight line is represented by the sides of a decagon by comparing the slopes and thus a chain code is obtained from this. In the last level, matching of an unknown chain code is done with the chain codes from the database with an accuracy rate of 89.3%. This simple yet effective code is very useful for e-learning of `Bharatanatyam´ dance. It is a fast and effective way to spread the dance form world-wide.