基于交通视频的能见度估计研究

发布时间：2018-01-23 02:22

本文关键词： 智能交通系统大气能见度暗通道先验交通监控视频　出处：《上海交通大学》2014年硕士论文　论文类型：学位论文

【摘要】：由于大气能见度对交通中的人和车的行为有重要的影响,道路能见度的监测一直是智能交通系统的一个重要组成部分。恶劣的能见度状况不仅不利于人们的健康,而且容易引发交通事故,威胁人们的出行安全,因此对于低能见度的检测与预警尤为重要。传统的能见度测量方法主要包括人工目测方法和仪器测量方法。前者主观性强,精度差；后者操作复杂,成本高昂。因此,改进能见度测量技术具有重要的现实意义与应用价值。本文在对能见度相关的物理模型研究的基础上,讨论了不同应用场景下的基于图像理解的能见度估计方法,论述了其相关的测量原理,并提出了一个基于交通视频的能见度等级估计方法。具体的说,该方法首先从视频图像中提取出全局平均梯度和暗通道比率特征,并由此建立能见度等级的估计模型,最后用一个直方图校正模块和一个晴天检测模块对估计结果进行修正。本文通过大量的实验来验证了该方法的有效性。此外,本文还自主研发了一个道路环境信息的集成平台,以更全面地反映道路环境信息。其集成的信息包括能见度等级、温度以及湿度。该平台主要包含3个方面的功能：第一,根据输入的道路监控视频估计大气能见度等级；第二,通过USB接口读取温湿度传感器采集的温度和湿度信息；第三,将获取的温度、湿度以及能见度等信息通过视频水印的方式进行集成与可视化。
[Abstract]:Atmospheric visibility has an important impact on the behavior of people and cars in traffic. Visibility monitoring is an important part of intelligent transportation system. Poor visibility is not only harmful to people's health, but also easy to cause traffic accidents and threaten people's travel safety. Therefore, it is very important for low visibility detection and early warning. The traditional visibility measurement methods mainly include manual visual measurement method and instrument measurement method. The former has strong subjectivity and poor precision. The latter is complex in operation and high in cost. Therefore, improving visibility measurement technology has important practical significance and application value. Based on the study of the physical model of visibility correlation, this paper discusses the visibility estimation methods based on image understanding in different application scenarios, and discusses the related measurement principles. A visibility level estimation method based on traffic video is proposed. In detail, the global average gradient and dark channel ratio feature are extracted from the video images. Finally, a histogram correction module and a sunny detection module are used to correct the estimation results. A large number of experiments are carried out to verify the effectiveness of the proposed method. In addition, this paper also developed a road environment information integration platform to reflect the road environment information more comprehensively. The integrated information includes visibility level. Temperature and humidity. The platform mainly includes three functions: firstly, the visibility level of atmosphere is estimated according to the input video of road surveillance; Secondly, the temperature and humidity information collected by the temperature and humidity sensor is read through the USB interface. Thirdly, the information of temperature, humidity and visibility is integrated and visualized by video watermarking.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;TP391.41

【参考文献】