运动序列图象中人体关节位置快速定位与应用

为解决运动图像中人体关节自动快速定位问题 ,提出了一种体育运动序列图象中的人体关节定位的快速算法 ,从运动序列图象中提取运动信息。运用序列图象前帧图象中关节位置的先验知识对运动关节在前后帧中进行块匹配 ,得到后帧图像中人体关节的位置 ,继而计算出科学训练中教练员所关心的一些运动参数。其中许多前提条件和假设都是基于高速摄像机或摄影机拍摄的图象的。在分析过程中 ,首帧图像的关节位置由人工输入 ,作为先验知识 ,以后的分析过程由计算机自动进行 ,分析结果迅速准确。作为应用实例 ,对棒球运动和跳高运动的技术录象资料的有关参数进行了提取和分析。

Quick Fixation of Human Body Joint Positions in Sports Serial Images and its Application

In order to find out the way to solve the problem of fixing the positions of human body joints quickly and automatically, the method of quick calculation to fix the positions of human body joints in sports serial images was advanced. The sports information could be obtained from the sports serial images. By applying the presupposition and the contrast images of the front views of joint positions in the serial images and matching the front views and back views of the exercise joints, the exact positions of human body joints in back views were obtained. Thus, the parameters of scientific training, which the coaches are most interested in, were obtained. Many of the presuppositions and suppositions were based on the pictures taken by high speed pickup cameras or cameras. During the process of analysis, the joint positions in the first images were input artificially which are regarded as the presupposition, and the following analysis of the joint positions were done by computers. The results of the analysis were achieved quickly and accurately. Using them as the practical examples of application the parameters of the baseball and high jump technical video data were collected and analyzed.