Real-Time Hand Motion Recognition: A Robust_x000D_ Low-Cost Approach

Abstract

This study presents a robust, low-cost hand motion_x000D_ recognition approach designed to run on low-end computer_x000D_ systems. Our method detects and tracks hand region using realtime images obtained from a low-resolution camera (i.e. webcam)_x000D_ and is not depended on any training or calibration and is not_x000D_ required any special camera apparatus or selectors. The proposed_x000D_ system involves several image processing techniques such as_x000D_ background subtraction, face detection, skin colour detection and_x000D_ template matching. The experimental results show promising_x000D_ performance under various conditions. The method has a wide_x000D_ range of applications where more natural ways of interaction_x000D_ required, such as virtual reality applications, assistive technologies_x000D_ and simulation.