无人机视频中道路交叉口车辆检测与跟踪  

A Vehicle Detection and Tracking Model for Road Intersections Based on UAV Video

作　　者：胡继华 程智锋[2] 钟洪桢[3] 靖泽昊 张力越 HU Jihua;CHENG Zhifeng;ZHONG Hongzhen;JING Zehao;ZHANG Liyue(Public Labs Center,Sun Yet-sen University,Guangzhou 510006,China;Guangdong Provincial Key Laboratory of Intelligent Transportation System,Sun Yet-sen University,Guangzhou 510006,China;School of Data and Computer Science,Sun Yet-sen University,Guangzhou 510006,China)

机构地区：[1]中山大学公共实验教学中心,广州510006 [2]中山大学广东省智能交通重点实验室,广州510006 [3]中山大学数据科学与计算机学院,广州510006

出　　处：《实验室研究与探索》2019年第5期28-31,106,共5页Research and Exploration In Laboratory

基　　金：国家自然科学基金项目(41271181);广东省科技计划项目(2016A010101015)

摘　　要：提出了一种基于无人机视频的交叉口车辆检测和跟踪方法,以道路交叉口行车区域为检测区域,将车辆检测和跟踪分成独立的两个阶段,并使用背景差法检测车辆,接着使用置信度指标进行车辆跟踪。该方法使用广州大学城两个道路交叉口的视频进行了验证,车辆检测和跟踪结果的精度都达到94. 49%以上,表明该方法准确可靠。基于无人机视频的车辆检测跟踪方法具有实施方便、快速和适用范围广等特点,为道路交叉口车流量调查提供了新方法,可以用于道路交通的实验教学、科研及生产等领域。In this paper,a method of vehicle detection and tracking based on UAV video is proposed. The method is applied to the videos of two road junctions in Guangzhou University City,the accuracy of vehicle detection and tracking results are all more than 94. 49%,which means the method is accurate and reliable. The vehicle detection and tracking method based on drone video has the characteristics of convenient implementation,rapidity and wide application range.It provides a new method for traffic survey of road intersections,the method is suitable for the fields of experimental teaching,scientific research and planning of road traffic.

关 键 词：智能交通 车辆跟踪 车辆检测 道路交叉口 无人机视频 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]