Independent Hand Gesture Recognition

摘要

The system we designed for enhanced user interface abilities allows user to communicate with computers by hand gesture. The system consists of three basic parts: hand target detection, feature vector extraction, and gesture recognition. For detecting the hand target in real time, a scale-judged method based on frame difference technique is developed. Polygon approximation of hand and an approach based on CC(convex-concave) representation for hand gesture is used to feature vector extraction. And based on the feature , both a graph-matching method based on sub-part of hand and a neutral network are used in the part of recognition . This paper describes our work on design and implementation for the system.