基于边缘计算的视频监控框架

Edge computing based video surveillance framework

为降低云平台视频监控系统视频传输时延、视频分析处理开销和存储成本,提出一种基于边缘计算的视频监控框架。根据边缘节点地理位置靠近视频源的特点,将监控视频流的分析预处理工作迁移到边缘节点完成;设计边缘节点帧过滤算法,通过对视频流的目标识别,筛选出有效视频帧进行上传;在此基础上,设计边缘节点任务调度算法,提高视频流实时性和资源利用率。实验结果表明,该框架能够有效降低视频传输开销和视频分析处理开销。

To reduce transmission delay,video processing costs and video storage space of the cloud platform video surveillance system,a video surveillance framework based on edge computing was proposed.Based on the characteristics of the ECN’s location near the source of surveillance,the analysis and preprocessing of the monitored video stream was offloaded to the ECN.The Frame-Selection algorithm was designed to select the effective video frames for uploading through the object detection of the video stream.On this basis,the ECN controller scheduling algorithm was designed to improve the timeliness and resource utilization of video streaming.Experimental results show that the proposed framework can significantly reduce unnecessary video traffic and the consumption of computing resources.