Essential Human Body Points Tracking Using Kalman Filter

Abstract— Human body tracking has a wide spectrum of applications. Examples include video surveillance, motion and activity capture for medical analysis, just to name a few. In this paper, a human skeleton model is proposed for human body tracking in video surveillance system. The skeleton model is based on 8 important body points namely; head, neck, shoulder, pelvis, knees and, ankles. Basically two techniques are used to locate these points. The first technique uses gait study to find the knee and ankle points. The self occlusion of human legs itself was overcome using the gait information. The gait information is then used to overcome self-occlusion of human legs. The second technique involves Kalman filter to track the remaining proposed body points.