运动图像序列的人体三维运动骨架重建

3D Human Skeleton Reconstruction from Motion Image Sequence

在计算机视觉领域 ,人体运动分析的研究具有相当广泛的应用前景 .由于人体运动的复杂性 ,已有的研究方法都对研究对象——人体 ,加上了许多限制条件 .文中提出了一种新的方法来获得各类人体运动的信息 .文中主要论述了该方法中对人体三维运动骨架的重建部分 ,其基本思想是在摄像机定标的基础上 ,应用三维人体模型知识和运动连续性依次建立每个图像的人体运动骨架 .

In the domain of computer vision, there is a very wide application for the study of human motion analysis. Due to complexity of human motion, existing methods lay much limitations on human body. This paper proposes a new approach to obtain the information of human motion. It focuses on the three dimensional human motion skeleton reconstruction. The main idea is using three dimensional human model knowledge and motion frame coherence on the basis of camera calibration. The final result is presented at the end of the paper.