Movement Tracking in Real-Time Hand Gesture Recognition

Abstract

To translate the gesture performed by the user in a video sequence into meaningful symbols/commands, feature extraction is the first and most crucial step in such systems which measures the detected hand positions and its movement track. We propose an efficient approach based on inter-frame difference (IDF) to handle the hand movement tracking, which is shown to be more robust in the accuracy aspect compared to skin-color based approaches. Computational efficiency is another attractive property that our approach greatly improves the processing frame rate to fulfil the demand of a real-time hand gesture recognition system.