A hand gesture interaction system based on Kinect

Weiqing Li, Zehui Lu, Shihong Shen

School of Computer Science& Engineering, Nanjing University of Science& Technology, Nanjing, China, 210094

We introduce a hand gesture interaction system using Kinect, which takes advantage of real-time dynamic motion capture, image recognition and so on, so that people can interact with computer by natural hand gestures. Five kinds of gesture are defined and can be recognized by the system. Kinect-based hand movements and gesture recognition algorithm is studied. A method for hand area image segmentation from the depth map is proposed, using the information of hand joints in skeleton map. We realized a hand gesture recognition algorithm with SVM and tested its stability and robustness. Finally, experimental results verified the feasibility of the algorithm, and a hand gesture interactive demonstration is implemented.