HMM Static Hand Gesture Recognition Based on Combination of Shape Features and Wavelet Texture Features

Abstract

Gesture recognition is one of the key technologies in the field of computer vision, and hand gesture recognition can be divided into static hand gesture recognition and the dynamic hand gesture recognition. This paper presents a new static gesture recognition algorithm based on hidden markov model. It uses two kinds of new shape features, the specific angle shape entropy feature and the upper side contour feature. They are firstly used for parameters training of hidden makov model, and then identify gesture categories hierarchically. In order to further improve the recognition effect for those small shape differences gesture, this paper adopts wavelet texture energy feature which can reflect the internal details of the gesture image, and makes the final correction estimation based on minimum total error probability. The experimental results show that the method has good recognition effects for gestures no matter the shape differences are big or not, and it has good real time performance as well.