摘要
对比现有典型关键帧提取算法存在的缺陷,提出了一种类模糊C均值聚类的关键帧提取算法.该算法预先设定一个最初聚类中心,从而有效地减少了聚类算法的迭代次数;并使用视频帧序列的时序特征来对FCM算法进行限定,提高了聚类效率.实验结果表明,使用该算法提取的关键帧不仅可以充分表达出视频的主要内容,而且还可以根据内容的变化提取出适当数量的关键帧.
Contrast the current typical algorithm to extract key frames,a new algorithm was proposed,which was named Simu-FCM Clustering.In this algorithm,the iterative times to cluster were reduced,since that a initial clustering center was given in advance;and the clustering efficiency was improved since the algorithm of FCM was restricted using the time sequence character of the video frame frequency.The experiment result shows that,the key frames extracted by using this algorithm could adequately express the prima...
出处
《微电子学与计算机》
CSCD
北大核心
2009年第2期89-92,共4页
Microelectronics & Computer
基金
国家自然科学基金项目(60572080,60772151)
关键词
视频
关键帧
模糊C均值
聚类
video
key frames
fuzzy C-means
clustering
