摘要
体育视频中目标标识在机器视觉等领域有广泛的应用而基于传统特征的目标标识运算量大、鲁棒性差,无法满足视频标注的实时性的要求。本文基于SURF特征对篮球视频中的目标进行标识,首先利用SURF算法获取特征点,然后利用最近邻匹配法找到匹配点对,最后利用所求值得到标识后的图像。实验表明该算法能满足准确性和实时性的要求,在目标出现一定的旋转状况下仍能实现准确的标识目标。
Marking the object in sports video is widely used in machine vision, but, marking based on traditional features is heavy in computation and poor in robustness, such that it can not fulfill real-time demand in video annotations .This paper mark the object in basketball video basing on SURF features. In the first place, the SURF algorithm is used to get the feature descriptor, and then nearest neighbor matching method is used to find the matching points, finally obtain the marked figure. The results of experiment show that this algorithm can meet the requirements of accuracy and real-time. Though the object is rotated, it can be marked by the new algorithm.
出处
《科技信息》
2011年第1期I0062-I0063,共2页
Science & Technology Information
