The simulated mouse method based on dynamic hand gesture recognition

Abstract

The existing simulated mouses are mostly based on the hand tracking technique, which depend on the data gloves, remote controllers or other equipment, leading to high costs and severe user limits. In this paper, we develop a new kind of simulated mouses by employing dynamic hand gesture recognition in which the palm node is detected by a Kinect camera. According to the movements of users in real world, we can realize the virtual control of the computer mouse. In the Kinect, since the palm node is defined as a skeleton one, those complicated operations, such as the motion detection, target classification and gesture tracking which are used in the hand tracking, are all omitted. Furthermore, our proposed simulated mouse does not rely on any other auxiliary equipment, relieving the limits on users. Experimental results show that under the same testing environment, the proposed simulated mouse can achieve the gesture recognition rate by 99.2%, 21.4% higher than that of the hand tracking one.