Visual-based interface using hand gesture recognition and object tracking

Abstract

This paper presents a novel approach for human-machine interface using visual-based communication. Scene analysis, object recognition and tracking, together with gesture detection and classification, are employed to design and implement a system for real time interaction. A new paradigm is proposed for efficient selection of hand gestures that can be used in other gesture-based environments aiming at interactive multimedia. An inexpensive web camera and readily available tools (such as an ordi­nary pen) have been used in this approach to render the system cost effective and suit­able for general and home-based multimedia applications. Focus is placed on real time applications with a fast decision function for gesture classification, and a model-based approach for orientation detection. Experimental results demonstrate the efficacy of the proposed approach when tested using different performance criteria