Human pose recognition using Kinect and rule-based system

Kinect - a motion sensing device that developed by Microsoft has opened up a lots of interesting application in various fields. In this paper, we studied the effectiveness of Kinect and a rule-based system to identify a series of human actions. First of all, a skeletal model with 20 joint positions is engaged as the human model. Secondly, we track the skeletal model and use the skeletal and colour information to build our rule-based system. Finally, we identify different human poses and action to perform different tasks in real time. Experimental results have shown the effectiveness of our proposed studied to classify different human pose and action.