摘要
为解决基于聚类算法的视频摘要中存在的需要预先设定聚类中心和聚类数目及选取的关键帧不具有代表性的问题,提出一种基于层次聚类与TextRank算法的静态视频摘要方法。使用层次聚类算法对视频帧进行聚类,利用TextRank算法选取候选关键帧集合,通过求解优化函数选择最终的关键帧生成视频摘要。实验结果表明,该方法生成的视频摘要能够比较全面准确地表达视频内容且冗余度低。
To solve the problem of pre-setting clustering center and the number of clusters in video summarization based on clustering algorithm,and the problem that the selected keyframe is not representative,a static video summary based on hierarchical clustering and TextRank algorithm was proposed. Hierarchical clustering algorithm was used to cluster video frames. TextRank algorithm was used to select candidate keyframes. The final keyframes were generated by solving optimization functions to gene- rate video summary. Experimental results show that the keyframe set extracted using this algorithm can express video content more comprehensively and accurately,and the redundancy is low.
作者
张璐
吕进来
ZHANG Lu;LYU Jin-lai(College of Information and Computer Science,Taiyuan University of Technology,Jinzhong 030600,China)
出处
《计算机工程与设计》
北大核心
2019年第7期1945-1949,共5页
Computer Engineering and Design
基金
山西省2013年科技攻关基金项目(20130321007-02)
关键词
视频摘要
视频帧分类
关键帧提取
层次聚类
TextRank算法
video summarization
video frames classification
keyframe extraction
hierarchical clustering
TextRank algorithm