| 视频分割与场景聚类算法研究 |
| |
| 引用本文： | 侯冠华,史萍.视频分割与场景聚类算法研究[J].中国传媒大学学报,2006,13(2):32-37. |
| |
| 作者姓名： | 侯冠华 史萍 |
| |
| 作者单位： | 国家知识产权局专利审查协作中心,北京,100083;中国传媒大学,信息工程学院,北京,100024 |
| |
| 摘 要： | 本文介绍了实现视频分割和场景聚类的算法.首先用X2直方图匹配法检测镜头切换,梯度法检测镜头淡入淡出;然后对镜头内非相邻帧间距离经过阈值判断提取关键帧;进一步基于镜头的关键帧集计算镜头间的相似度,对相似度大于阈值的镜头进行聚类;最后给出了典型实验结果,表明上述算法对视频分割和镜头聚类有较好的性能.
|
| 关 键 词： | 视频分割 关键帧提取 场景聚类 |
| 文章编号： | 1673-4793（2006）02-0032-06 |
| 收稿时间： | 2006-04-14 |
| 修稿时间： | 2006年4月14日 |
The research of video segmentation and scene clustering algorithms |
| |
| HOU Guang-hua,SHI Ping.The research of video segmentation and scene clustering algorithms[J].Journal of Communication University of China Science and TEchnology,2006,13(2):32-37. |
| |
| Authors: | HOU Guang-hua SHI Ping |
| |
| Abstract: | This paper introduced algorithms for video segmentation and scene clustering. The X2 histogram match and grads method is used for shot detection. A method of extracting shot frame sets based on distance of non adjacent frame is presented. Further more, the maximum similarity of key frame set as similarity of shots is computed and the scenes cluster according to the shot similarity. At the end of this paper, the experiment results is presented, which indicate that this algorithm has satisfied performance in correctness and computation speed. |
| |
| Keywords: | Video Segmentation Shot Frame Extraction Scene Clustering |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
