摘要
镜头检测在视频检索中起着关键作用。文章提出了一种基于信息论的镜头检测方法,该方法使用相邻帧间的交互信息量检测突变,使用相邻帧间的联合熵检测淡入/淡出镜头。试验表明:该方法简单、有效,且识别的准确率较高。
Shot detection is the most important issuse in the field of video retrieval.A shot detection method based on information theory is proposed.Mutual information of the neighbored frame is computed for cut change dectection.Joint entropy of neighbored frame is computed for detection fade in/fade out.The experiment shows that the approach is convenient and efficient ,especially with high reliability.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第30期54-56,共3页
Computer Engineering and Applications
基金
湖南省自然科学基金资助项目(编号:03JJY6022)
关键词
镜头检测
熵
交互信息量
shot change detection,entropy, mutual information