Abstract. Hand gesture recognition is regarded as an im-portant part of arti cial intelligence. A great efort was put into humancomputer interaction so that hand gesture recognition is gradually becoming a developed technology. In light of the utilization of mouse and keyboard, the increasing needs of human-computer interaction cannot be met; hindrance turns out to be increasingly genuine. In this paper, we reviewed previous investigations of vision-based gesture recognition and summarized their ndings. This paper compares the most common human-computer interaction products in recent years, which can be used to capture gesture data. Then we started with the classi cation of gestures and summarized the research of visual gesture recognition based on static and dynamic gestures. The gesture representations we summarized includes appearance-based and 3D model-based methods. We also introduced the applications of the two kinds of hand gestures recognition in the papers of recent years. A possible classi cation methods was put forward to improve the performance of gesture recognition. The goal of this paper is to summarize

the current technology and research results and compare the diferences and the advantage of di erent hand gesture recognition methods, which will contribute to the following research.

Keywords: Interaction products  Hand gesture recognition  Gesture representation  Application  Classi cation.