摘要
为快速有效地表征视频内容,提出一种视频关键帧抽取方法。结合Tsallis熵和Jensen距离计算相邻视频帧间差距,将视频分割为镜头,并根据镜头内视觉内容变化的多少将其分割为子镜头,最终抽取关键帧。同时提出视频帧间的距离度量标准,用于自适应地选取最优Tsallis熵指数。实验测试结果表明,该方法简单高效,对物体运体有较好的鉴别能力。
In order to summarize the main content of a video sequence quickly and effectively,this paper proposes a new method for extracting video key frames. The proposed technique makes use of Jensen-Tsallis distance to estimate the frame-by-frame distance betw een consecutive video images,to segment a video into shot,and into subshot either with or without large change of the video content. Video key frames are selected based on the subshot. It also proposes a measure for the difference between video frames to adaptively set an optimal index used for Tsallis entropy. Experimental result shows that the novel method is simple yet effective,and has better ability to identify the moving objects.
出处
《计算机工程》
CAS
CSCD
北大核心
2016年第2期278-282,共5页
Computer Engineering
基金
国家自然科学基金资助项目(61471261
61179067
U1333110)
