基于视频的路况能见度检测系统的设计与实现  

Visibility detection system based on road monitoring camera

作　　者：周庆逵[1] 陈钊正[1] 陈启美[1] 

机构地区：[1]南京大学电子科学与工程系,南京210093

出　　处：《电子测量技术》2009年第6期72-76,共5页Electronic Measurement Technology

基　　金：江苏省重大交通科研项目(07X06)

摘　　要：能见度涉及高速公路的安全行车,现有气象能见度仪本身样本空间受限,且价格十分昂贵而难以沿途布设,不能真实反映路段的能见度及其分布。为此,提出了基于路况监控视频的能见度检测系统的设计方案,阐述了视频能见度检测的基本原理,给出系统的总体框架,重点研究了能见度检测模块的实现:通过带有距离信息的目标图像块的选取、图像块的匹配、边缘特征的提取,空间亮度对比的计算、曲线拟合得到能见度信息。系统已在江苏省高速公路宁通、宁常路段试用,系统运行良好,检测结果与人眼观察误差符合国际规定(+10%),并具有较好的鲁棒性。Visibility is an important factor in the highway traffic. The meteorological visibility sensors used to detect fog and poor visibility are too expensive to set up along whole sections of road to reflect the true situation of road automates. Base on the principle of visibility detection, a realization approach is presented to measure atmospheric visibility based on traffic video-surveillance system. The system integrates target image htock selected, SAD, edge detection base on Wavelet Transformation, targets＇ contrast computation and curve fitting to get visibility distance. The system has been successfully applied in NingTong, NingChang road section of Jiangsu Province highway system, the results of detection are uniform to human perception （±10%） and meteorological sensors, and have good robustness.

关 键 词：能见度检测 模式识别 图像处理 边缘检测 曲线拟合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]