Inverse Kinematics and Extended Kalman Filter Based Motion Tracking of Human Limb

This paper addresses the problem of motion tracking of human skeleton system using non-invasive vision based sensors. The proposed approach combines synergistic paradigms of image processing, kinematics of rigid bodies and Extended Kalman Filtering scheme to estimate the motion of a human limb system. This approach solely depends on the measurement obtained from the vision sensors without involving any wearable or inertial sensors to measure the motion parameters. In this paper we propose fusion of two filtering schemes — the optical flow equations applied to raw images obtained from the Microsoft Kinect and extended Kalman filter for human skeleton considered as a system of kinematic linkages. The strategy proposed in this paper yields near optimal results as is demonstrated with the help of experiments performed using the Microsoft Kinect sensor and compared using accurate tracks obtained from 24-Camera Optitrack motion capture system.