摘要
随着视频监控系统的大量部署,产生了大量的监控视频数据,视频摘要技术应运而生。如何优化大量的视频摘要算法服务器的工作效率,高效的视频摘要任务调度方法成为关键。提出一种基于视频摘要任务执行时间预测模型的分布式任务调度方法。该方法对视频摘要处理过程中的监控视频数据块所需的服务器处理时间进行预测,从而合理调度任务,使服务器负载均衡,提高了云服务器的资源利用率,降低了视频摘要任务的分布式处理时间。
With the wide deployment of video surveillance systems, a large number of surveillance video data is produced. Video synopsis technology can get valid information in a fast and efficient way. Therefore, the optimization of the efficiency of a large number of video abstraction jobs and the video abstraction task scheduling method has become important factors. In this paper, a video abstraction task scheduling method based on the execution time prediction model in cloud computing platform is proposed. This method predicts the execution time of the video abstraction jobs in the computing platform based on the video data block＇s information and the server running state information. Then it schedules the video abstraction jobs appropriately so that it balances the server load, improves the level of resources utilization of the cloud server and reduces the distributed processing time of the video abstraction jobs.
出处
《计算机应用与软件》
2017年第7期7-10,共4页
Computer Applications and Software
基金
国家自然科学基金项目(61300013)
高等学校博士学科点专项科研基金项目(20130005120011)
关键词
视频摘要
任务调度
分布式计算
Video synopsis Task scheduling Distributed computing
