说明：  本文提出了有模型指导的三维人体运动跟踪框架，将一个多关节的圆台形状三维人体模型与多个视频图像中的外轮廓、边界、灰度和肤色特征进行匹配，使人体运动跟踪变成一个状态估计问题。并且，使用基于概率模型的粒子滤波算法来完成非线性、非高斯动态系统的状态估计。
(In this paper, the model has guided three-dimensional human motion tracking framework, will be more than one of the frustum of a cone shape of the joints of three-dimensional human body model with a number of video images of the outer contour, border, greyscale and color matching features, so that human motion tracking into a state estimation problem. In addition, the use of probabilistic model based particle filtering algorithm to complete the non-linear, non-Gaussian dynamic system state estimation.)