摘要
为了解决目前基于视频图像自动识别牌照进而推算动态OD的系统缺乏研究手段和工具的问题,应用Visual Ba-sic编程语言,通过组件对象编程技术,研究开发了基于微观仿真模型VISSIM的动态交通OD估计仿真试验平台,包括数据库系统、仿真模型系统以及数据采集处理系统。分别设计了最短路算法、车辆转向率法以及BP神经网络法以提高OD估计精度。上海陆家嘴区域的案例研究结果表明:车辆转向率法OD估计误差在5%以下,其计算效率和精度最优;在视频检测器满布状态下满足动态交通管理OD要求的视频牌照识别精度阈值为80%。
In order to solve the problem that there are lack of research tools and methods to estimate the dynamic OD based on video license plate automatic recognition, by using the programming language Visual Basic and component object programming technology, the dynamic OD estimation simulation platform which includes the database system, simulation model system, and data collecting and processing system was built based on micro traffic simulation model VISSIM.The shortest path algorithm, vehicle turn-ratio algorithm and BP neural network algorithm were put forward to improve the dynamic OD estimation precision. A case study indicates that （ 1 ） the efficiency and precision of vehicle turn-ratio algorithm is the best and its error is less than 5% ; （2） if the OD matrix meets dynamic traffic management, the lower bound of video license plate recognition precision is 80% under the full video detector layout condition.
出处
《公路交通科技》
CAS
CSCD
北大核心
2009年第8期130-134,共5页
Journal of Highway and Transportation Research and Development
基金
国家高技术研究发展计划(八六三计划)资助项目(2007AA11Z221)
上海自然科学基金资助项目(07ZR14120)
关键词
交通工程
动态OD
牌照识别
转向率法
微观仿真模型
视频检测器
traffic engineering
dynamic OD
license plate recognition
turn-ratio algorithm
micro simulation model
video detector
