摘要
针对篮球赛事视频分割问题,提出一种基于场地三分线的新分割算法。首先提取出帧图像的Canny边界,再对提取到的结果进行优化去杂;接着在优化过的边界中搜寻三分线并记录下三分线上点的坐标;然后根据这些点的坐标来判断三分线的弯曲方向,将其作为后续分割的依据。基于以上算法,以Microsoft Visual Studio 2010与Open CV2.3为开发平台,实现了一个篮球赛事视频镜头分割的原型系统,并通过该系统验证了新方法在正确检出和漏检方面的优势。虽然不是很明显,但是在误检方面的优势很显然。
In view of basketball tournament video segmentation, propose a new segmentation algorithm based on the three-point line. First, the Canny boundary of frame image is extracted and optimized, and then the three-point line has been searched with the coordinates of those points recorded which determine the bending direction of the three-point line and could be used as the basis for the subsequent split. By the development platform of Microsoft Visual Studio 2010 and OpenCV2.3, a prototype system of basketball tournament video shot segmentation has been achieved, which verifies that the new method＇ s advantage is not obvious in terms of being correctly detected and undetected,but it＇s notable in terms of false detection.
出处
《计算机技术与发展》
2015年第2期216-220,共5页
Computer Technology and Development
基金
上海市自然科学基金资助项目(11ZR1415200)
关键词
镜头分割
Canny边界
三分线
弯曲方向
shot segmentation
Canny border
three-point line
bending direction
