摘要
视频的子片段检测和分割是视频分析的一个重要步骤,提取视频中的关键信息能大大减少视频索引的数据量。基于此,简要介绍了传统的特征聚类技术,在聚类算法的基础上,提出一种聚类评估的方法,通过数值表示聚类的匹配程度,并对结果进行优化得到最佳聚类中心个数,保证了视频子片段检测的精度。实验证明该方法可较好地分割视频子片段。
Detection and segmentation of video sub-fragments is an important step in video analysis. Extracting key information from video can greatly reduce the amount of data indexed by video. Based on this, the traditional feature clustering technology is briefly introduced. On the basis of clustering algorithm, a clustering evaluation method is proposed. The matching degree of clustering is expressed by numerical value, and the optimal number of clustering centers is obtained by optimizing the results, which ensures the accuracy of video sub-fragment detection. Experiments show that this method can segment video sub-segments well.
作者
钟意
陈勇
Zhong Yi;Chen Yong(College of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China;Laboratory of Electrical Theory and New Technology, Chongqing University of Posts and Telecommunications, Chongqing 400065, China)
出处
《信息与电脑》
2019年第4期44-45,共2页
Information & Computer
关键词
视频子片段
特征聚类
中心个数
video sub-segment
feature clustering
number of centers
