一种新的基于对象的足球视频镜头分类方案  

A Novel Shot Type Classification Based on Object in Soccer Video

作　　者：周艺华[1] 曹元大[1] 张龙飞[1] 

机构地区：[1]北京理工大学信息科学技术学院计算机科学工程系,北京100081

出　　处：《计算机工程与应用》2005年第34期229-232,共4页Computer Engineering and Applications

摘　　要：论文提出了一种基于对象的足球视频镜头分类方案。首先对足球视频中的场地和运动员对象进行检测和分割,然后利用识别出的场地特征、运动员数目及运动员与场地比例等特征,对足球视频中的长距镜头、中距镜头、特写及其它类型的镜头进行分类。实验表明,该分类方案取得了良好的效果。This paper presents a novel shot type classification scheme based on object in soccer video.Firstly,Field and Athlete objects are detected and segmented.And then long shot,medium shot,close-up shot and other shots are classified by the features such as field area,number of athletes and the ratio between athlete and field.Experiment shows that the scheme performs good performance.

关 键 词：镜头分类 特征提取 视频摘要 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]