基于视频监控网络模型的行人搜索跟踪策略研究  

Pedestrian Tracking Search Strategy Based on Monitoring Topology Network Model

作　　者：蔺志宇 陈国良[1] LIN Zhiyu;CHEN Guoliang(School of Mechanical and Electronic Engineering,Wuhan University of Technology,Wuhan 430070,China)

机构地区：[1]武汉理工大学机电工程学院,湖北武汉430070

出　　处：《数字制造科学》2022年第4期275-281,共7页

基　　金：国家自然科学基金资助项目(61672396)

摘　　要：针对监控系统对目标行人进行搜索与跟踪时,常常面临监控点数量多、监控视频数据量大等挑战,并且由于监控视频在存储上的离散性和监控系统缺乏整合分析机制,完成该任务的过程缺乏自动化和智能化的问题,提出一种基于监控拓扑网络模型的目标行人搜索跟踪策略。首先将视频监控节点通过线性参考系统LRS坐标定位于拓扑地图中,构建视频监控网络模型并改进邻接表对模型进行存储;然后以该模型为基础,提出基于广度优先搜索算法BFS和深度优先搜索算法DFS的目标搜索跟踪策略;最后根据实际场景设计仿真实验,验证了搜索策略的可行性和有效性,实验证明通过该策略进行目标对象的搜索跟踪,综合成功率可达85%以上。When using the monitoring system to search and track the target pedestrian,it often faces many challenges such as a large number of monitoring points and video data,discrete storage of the data,and primitive analyzing capability.These problems limit the processing of the task from being intelligent and automated.In this paper,a pedestrian searching and tracking strategy based on monitoring topology network model is proposed.Firstly,the video monitoring node is located in the topology map through the linear reference system LRS coordinates.The adjacency table is then improved in the constructed video monitoring network model for storage.Then,a target search and tracking strategy based on BFS and DFS is proposed;According to the actual scene,the feasibility and effectiveness of the developed searching strategy has been verified by simulation.The results demonstrate that the comprehensive success rate of the target object searching and tracking can reach more than 80%.

关 键 词：视频监控系统 拓扑网络 邻接表 广度优先搜索 深度优先搜索 

分 类 号：TN948.6[电子电信—信号与信息处理] TP18[电子电信—信息与通信工程]