面向人车路环境的视频检测技术应用研究  

Human-vehicle-road-environment-oriented study on the application of video detection technology

作　　者：曹宇 何赏璐 陈新[1] CAO Yu;HE Shanglu;CHEN Xin(School of Automation,Nanjing University of Science&Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学自动化学院,江苏南京210094

出　　处：《交通科学与工程》2023年第3期98-109,共12页Journal of Transport Science and Engineering

基　　金：国家重点研发计划项目(2019YFE0123800);国家自然科学基金(52102380);江苏省自然科学基金(BK20180486);中国博士后科学基金(2018M642257和2021T140325);中央高校基本科研业务费专项资金(30920021140);中国交建2019年重大科技专项(2019-ZJKJ-ZDZX02)。

摘　　要：为全面了解视频检测算法在道路智能交通系统中的应用,深入探索该技术在未来的发展趋势,利用关键词进行分析,从检测对象的角度将该类算法分为路与环境、车、道路参与者三类,并对其进行对比、分析、归纳,总结其在道路交通不同应用场景中的应用现状;分类对比不同对象检测所采用的核心图像处理算法。从算法、检测对象、应用场景三个方面系统地分析和总结了视频检测技术的发展趋势和关键技术难点。研究结果表明:视频检测技术在道路智能交通系统中发挥了重要作用,应用广泛。未来其将有望在深度学习算法、复杂背景、多样化对象等方向深度发展。In order to comprehensively understand the application of video detection algorithms in road intelligent transportation systems and delve into the future development trends of this technology,a thorough exploration was conducted.Employing keyword analysis,the algorithms in this category were classified into three groups based on the perspective of detection objects:road and environment,vehicles,and road participants.A comprehensive comparison,analysis,and summarization of these algorithms were performed,along with a summary of their current application status in different road traffic scenarios.The core image processing algorithms utilized for various object detection tasks were categorized and compared.The development trend and key technical difficulties of video detection technology are systematically analyzed and summarized from three aspects:algorithms,detection objects and application scenarios.The research results show that video detection technology plays an important role in road intelligent traffic systems and is widely used.In the future,it is expected to further develop in the directions of deep learning algorithms,complex backgrounds,diverse objects.

关 键 词：交通工程 视频检测技术 智能交通系统 图像处理算法 深度学习 

分 类 号：U495[交通运输工程—交通运输规划与管理]