Kinematic Simulation of Human Motion using LifeMOD interfaced with Kinect

Abstract

Due to the high accuracy of the measured motion data, optical based motion capture has been used in many fields, such as sports analysis, computer vision, virtual reality and so on. However, there are also disadvantages of the expensive infrared cameras, cumbersome system construction, and the need to attach markers to anatomically defined position of the human body. Therefore, the motion capture system which based on marker-less and multiple views is a good alternative to intrusive marker system. In this study, double Kinects were used to collect marker-less human motion data which were then mapped to a LifeMOD human model for human motion simulation. Two Kinects were used to obtain the human motion skeleton from different perspectives, and the optimal skeleton was obtained by solving the optimization problem. Four key parameters (age, height, weight and waist circumference) of the subjects were entered and the human body semantic parameters were calculated by regression equations. The positions of the joints obtained from two Kinects were converted to the marker positions on the skin surface of the LifeMOD human model. Finally, the kinematic simulation of human motion can be implemented using LifeMOD interfaced with Kinect.