基于视频的三维人体运动跟踪系统的设计与实现

Design and Realization of Vision-Based 3D Human Body Motion Tracking System

在优化粒子滤波跟踪框架下,设计并实现了一个结合多种图像特征、在多摄像机环境下跟踪人体运动的三维人体运动跟踪系统.通过定义三维人体模型、摄像机模型以及观测似然模型,得到跟踪所需目标函数,并使用优化粒子滤波算法进行求解.实验结果表明,该系统能够对人体运动进行准确的跟踪和三维重建,可应用于体育运动分析和动画制作等领域.

A video-based 3D human body motion tracking system is developed under the optimization particle filter framework, which combines multiple cues in multiple cameras environment. By defining the 3D body model, camera model and image likelihood model, we can define the tracking object function and resolve it using optimized particle filter. The experiments show that tracking and reconstructing results made by our system are accurate and effective. This system can be applied to several areas such as sporting motion analysis and human animation.