视频数据的时间序列挖掘方法研究

不同类型的视频具有不同的内容和不同的内容组织结构.目前由于安全防护问题的升温,综合视频监控系统和交通监控系统引起了较多的关注.但是并没有文献对这一类型视频数据的挖掘提出一个较完整的挖掘框架.文中针对监控视频等具有相对固定背景特征视频数据的视频运动挖掘问题,提出了在跟踪视频对象运动的基础上,提取视频对象的运动轨迹,结合时间特性形成时间序列数据,并进一步对视频数据进行相似性查询和聚类挖掘的系统模型.

Research on Time Series Based Video Mining Techniques

The content and the structure of the video vary with the video type.The developments of complex video surveillance systems and traffic monitoring systems have recently captured the interest of many researchers due to the increasing safety and security concerns.But there is no framework for this type video mining.Focusing on the video motion mining,this paper proposed a video mining system using video object tracking techniques to extract motion in trajectories in video.