基于目标序列分析的智能监控多目标自动检测跟踪  

Multiple Objects Detection and Tracking in Intelligent Supervisory Based on Object Sequence Analysis

作　　者：刘李漫[1] 张治国 Liu Liman;Zhang Zhiguo(School of Biomedical Engineering,South-Central University for Nationalities,Wuhan 430074,China;School of Automation,Huazhong University of Science and Technology,Wuhan 430074,China)

机构地区：[1]中南民族大学生物医学工程学院,武汉430074 [2]华中科技大学自动化学院,武汉430074

出　　处：《中南民族大学学报（自然科学版）》2018年第4期130-136,共7页Journal of South-Central University for Nationalities：Natural Science Edition

基　　金：国家自然科学基金资助项目(61305044);中央高校基本科研业务专项资金资助项目(CZY17011)

摘　　要：针对道路交通监控场景,提出并实现了一个多目标自动检测与跟踪系统.首先通过高斯混合模型的运动信息提取方法得到目标运动信息,并分析了目标信息的连续多帧历史信息,估计目标区域信息在连续多帧中的统一性,进而通过信息融合的方式得到了目标检测结果.然后提出了监控场景下多目标的跟踪与管理策略,根据目标检测结果,对场景中的多个目标同时进行跟踪.最后,根据目标的不同状态将目标划分为新出现目标、被更新目标和被跟踪目标等不同的类别,提高目标跟踪的准确性.A multiple object tracking system is developed to deal with the road traffic monitoring.Firstly the object information is extracted by GMM model method and then the objects are detected by analyzing the sequence information of objects.Next the extracted objects are tracked by information fusion.Then a multiple object management strategy is proposed to handle the multiple objects tracked by the object detection.At last the objects are divided as the new object list,the updating object list and the tracking object list by the different states to improve the accuracy of object tracking.

关 键 词：多目标跟踪 智能监控 目标检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]