An efficient approach for computer vision based human motion capture

Abstract

In this paper, we present a novel computer vision based human motion capture approach by human body reconstruction process and energy function minimizing. After analyzing 3D human model in detail, we conduct human motion capturing by four steps, which are 1) Capturing the video, 2) Recognizing human feature points, 3) Tracking the feature points, and 4) Representing the motion movement. To test the effectiveness of the proposed approach, we conduct experiments on HumanEva dataset under four metrics. Experimental results show that our approach can capture human motion precisely. © 2013 Springer-Verlag.