Online dynamic hand gesture recognition with multiple cues

Abstract

In order to solve the generalization performance and complex background problems of hand gesture recognition, online dynamic hand recognition with multiple cues is proposed in this paper. The disturbance caused by complex background is reduced by motion detection. As a result of skin color's cluster characteristic, the online skin classifier is constructed by Multi-Gaussian model. The static hand recognition is completed with geometric features. An affine model is adopted for motion displacement estimation for hand tracking. The experimental results show that our method is robust and real-time, and is able to adapt to the complex background.