A Simple Manner of Dynamic Gesture Recognition Based on Kinect

The paper develops a simple method of dynamic gesture recognition based on the Kinect which is a new sensor from Microsoft in the environment of VS2010 combining Kinect for Windows SDK v1.8. Kinect sensors can track human bodies within their effective scope in real-time and obtain the depth of the corresponding information and bones at the same time. Firstly, we separated each dynamic hand gesture to be a combination of several micro-gestures which were from predefined eight micro-gestures with eight different directions. Then Dynamic Time Warping algorithm was adopted to find the most similar template gesture, then the system realized corresponding control instruction of the matched template gesture, and consequently, we can achieve somatosensory interaction by recognizing dynamic hand gestures, and the total recognition rate isn't less than 90%.