Recognition of hand gesture using hidden Markov model

Abstract

In this paper we proposed a recognition system for hand gesture in 3D environment by using only a single camera. For calculating the relative motion towards the camera, generally a depth sensing device is needed. In order to remove that, we proposed an approach of using the change of the area of the hand in input image. Using skin color; we detect the hand from the input image sequences and then we process the data for feature extraction. Three features are proposed for effectively recognize the gesture by our system. These are orientation, area and angle of the palm. As we proposed our system for dynamic gesture, Hidden Markov Model is utilized to recognize the gesture. In our lab environment our proposed system shows very promising result and we were able to achieve about 80.67% recognition rate on average. The system that we proposed will not only help to recognize the gesture of hand accurately but also lessen the cost for implementing this kind of system because of using minimal number of hardware.