基于小波变换的视频能见度检测算法研究与实现  

Video visibility detection algorithm based on wavelet transformation

作　　者：陈钊正[1] 周庆逵[1] 陈启美[1] 

机构地区：[1]南京大学通信技术研究所,南京210093

出　　处：《仪器仪表学报》2010年第1期92-98,共7页Chinese Journal of Scientific Instrument

基　　金：中国交通部科技攻关项目(2004-353-352-04);江苏交通厅科研基金项目(06x05;07x03)资助

摘　　要：由于传统能见度仪价格昂贵、架设困难、不能给出道路能见度分布,提出了基于小波变换的视频能见度检测算法。利用摄像机自标定模型恢复图像距离信息,采用小波变换提取图像边缘特征点,使用能见度检测模型提取符合人眼的特征,结合曲线拟合,给出能见度变化曲线。该算法及其系统已在江苏宁常镇溧高速公路上进行试点使用,其结果与人眼及能见度仪检测误差在+10%之内,符合相关国际及国家标准。Traditional visibility measurement sensor is expensive, hard to install and can＇t give road visibility distribution. This paper presents a method of video visibility detection based on wavelet transform. Camera self-calibration is used to recover the distance information from images and wavelet transform is used to get the points that belong to the edges of the target objects; then a visibility detection model is used to get the feature that is in accordance with human eyes; finally, with curve fitting, the visibility changing curve is obtained. Now the system based on the proposed model and algorithm has been used on Ning-Chang and Zhen-Li high-way; the application results of the system are consistent with those from human eyes and visibility measurement sensors, and the error between them is with ＋10%, which conforms to national and foreign standards.

关 键 词：小波变换 摄像机自标定 能见度检测模型 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]