摘要
随着体育竞技水平的提高，体育训练的要求也不断提高。为了适应这种现状，计算机视觉技术被应用到体育训练中，因为机器视觉比人眼有更好的准确性和记忆性，能够快速地捕捉运动目标，并且能够记录目标的各种运动数据。对体育运动视频中的非刚性目标进行跟踪，提出了基于均值移位算法与彩色直方图算法的跟踪方法，不仅能够对目标的位置进行有效的跟踪估计，同时对目标的形状也能很好地进行描绘，解决了体育运动视频中非刚性目标形状复杂不易跟踪的问题。给出了实例验证，表明所运用方法的有效性、自适应性。
Abstract
Sports training require continuous improvement with the increased level of sports competition. Therefore, computer vision technology is adopted. Machine vision has better accuracy and memory than human vision. It can quickly capture the moving target and record the movement data. In this paper, we research on the object tracking in sports video, and a tracking method based on mean-shift algorithm combined with color-histogram is proposed. The new method can simultaneously estimate the position of target object and the approximate shape. So it has excellent performance in tracking non-rigid object in sports video. Finally, some tests are brought out and the test results dem- onstrate the effectiveness, adaptability of the new method.
NETL
NSTL
维普
万方数据