曹月花+罗文广+蓝红莉+赵晓东

摘 要： 针对不同光照、雨雪天、大雾天等恶劣环境下道路车道线识别难的问题，研究其识别算法。弱光照下，通过面积算子选取合适阈值进而实现边缘增强；夜间光照更弱的情况下，通过改进SUSAN算子实现夜间昏暗道路的边缘增强，解决了道路信息模糊、周围因素影响等造成的噪声干扰；强光照下，通过强光滤波算法可以消除伪边缘干扰，实现车道线的准确识别。雨雪天下，道路积水造成路面复杂，不能准确找到车道线特征点，通过建立路面积水反射模型，去除光的反射影响，增强路面和车道信息对比度，实现车道线信息的准确提取。大雾天，道路信息模糊不清，通过逆透视原理、差值分离建立道路识别模型，对道路特征进行加强分析，增强车道信息，提高识别效果。实验验证了所提出和改进的复杂环境下道路车道线识别算法的有效性，并且具有较强的鲁棒性和抗扰能力，可应用到智能交通系统中。

关键词： 车道线； 车道线信息提取； 逆透视变换； 边缘增强

中图分类号： TN912.34?34； TP391.41 文献标识码： A 文章编号： 1004?373X（2017）14?0109?05

Abstract： It is difficult to recognize the lane line on road in the severe environments such as different illumination， rain day， snow day， fog day， etc， therefore a lane line identification algorithm is studied. Under the condition of weak illumination， the appropriate threshold is selected by means of area operator to realize the edge enhancement. Under the condition of weaker illumination at night， the SUSAN operator is improved to realize the edge enhancement of the dark road at night， and eliminate the noise interference caused by fuzzy road information and surrounding factors. Under the condition of strong light， the high light filtering algorithm can remove the pseudo edge interference， and realize the accurate recognition of the lane line. Under the conditions of snow or rain day， since the feature point of the lane line can′t be identified accurately because of the complex road conditions caused by road water?logging， the reflection model of road water?logging was established to eliminate the influence of light reflection， enhance the contrast of roads and lanes information， and extract the lane line information accurately. Under the condition of fog day， because the road information is blur， the road recognition model based on inverse perspective principle and difference separation was established to perform the enhancement analysis for road feature， enhance the lane information， and improve the recognition effect. The experimental results verify that the road lane line recognition algorithm is effective in complex environment， has strong robustness and anti?interference ability， and can be applied to the intelligent transportation system.

Keywords： lane line； lane line information extraction； inverse perspective transformation； edge enhancement

0 引 言

21世纪是一个科技和信息技术飞速发展的时代，直接带动交通运输业的发展。科技进步直接或间接地使劳动过程发生变化，促进交通行业经济增长，但是交通安全隐患也越来严重[1]。道路车道线识别是汽车安全辅助系统中最基本和核心的功能，能为导航提供精确的参照，在划分道路区域、安全驾驶和碰撞预警等方面起到了重要的作用。……