基于仿射传播聚类的视频关键帧提取方法  

VIDEO KEY FRAME EXTRACTION USING AFFINITY PROPAGATION CLUSTERING

作　　者：谢晓能[1] 吴飞[2] 

机构地区：[1]杭州广播电视大学,浙江杭州310012 [2]浙江大学计算机科学与技术学院,浙江杭州310027

出　　处：《计算机应用与软件》2009年第5期28-30,共3页Computer Applications and Software

基　　金：国家自然科学基金项目(60272031);浙江省自然科学基金(M603202)

摘　　要：关键帧提取是基于内容的视频摘要生成中的一个重要技术。首次引入仿射传播聚类方法来提取视频关键帧。该方法结合两个连续图像帧的颜色直方图交,通过消息传递,实现数据点的自动聚类。并与k-means和SVC(support vector clustering)算法的关键帧提取方法进行了比较。实验结果表明,AP(Affinity Propagation)聚类的关键帧提取速度快,准确性高,生成的视频摘要具有良好的压缩率和内容涵盖率。Key frame extraction has been recognized as an important technique in content-based video summarization generation. In this paper, affinity propagation clustering is introduced for the first time in key frame extraction, this new approach clusters data points automatically through messages transmission based on combining colour histogram crosses of two serial image frames. It has been compared with the key frame extraction methods of k-means and support vector clustering algorithm, experimental results show that it can quickly and accurately extract key frame of AP cluster, and the video summarization generated has fine compression rate and content coverage rate.

关 键 词：视频摘要 关键帧提取 仿射传播聚类 

分 类 号：TP311.134[自动化与计算机技术—计算机软件与理论] TP391[自动化与计算机技术—计算机科学与技术]