改进的光照不变道路检测算法

发布时间：2018-04-08 10:40

本文选题：智能交通　切入点：道路检测　出处：《交通运输系统工程与信息》2017年05期

【摘要】：针对大多数道路检测方法存在光照变化敏感,阴影导致误检、漏检等问题,提出了一种改进的光照不变道路检测算法.首先将道路图像RGB空间转换为几何均值对数色度空间;然后根据Shannon熵确定相机轴标定角θ,利用Chebyshev理论去除θ奇异值,得到光照无关图Iθ;其次通过随机抽样方法提取道路样本点,包括道路基准样本点和道路参考样本点;最后建立道路置信区间分类器,将道路从背景中分离出来.实验结果表明,该算法能很好地消除光照变化和阴影对道路检测的影响,检测精度高,能满足实际道路检测实时性要求.
[Abstract]:An improved illumination invariant road detection algorithm is proposed to solve the problems of sensitive illumination change, false detection caused by shadow and missed detection in most road detection methods.Firstly, the road image RGB space is transformed into the geometric mean logarithmic chrominance space, then the camera axis calibration angle 胃 is determined according to the Shannon entropy, the singular value of 胃 is removed by Chebyshev theory, and the illumination independent graph I 胃 is obtained. Secondly, the road sample points are extracted by random sampling method.Finally, a road confidence interval classifier is established to separate the road from the background.The experimental results show that the algorithm can eliminate the influence of illumination and shadow on road detection, and the accuracy of detection is high, and it can meet the real time requirement of road detection.
【作者单位】：长安大学电子与控制工程学院;江苏省交通运输与安全保障重点实验室;
【基金】：国家自然科学基金(61603057) 陕西省科技工业攻关项目(2015GY033) 江苏省交通运输与安全保障重点建设实验室开放基金资助(TTS2015-04)~~
【分类号】：TP391.41;U463.6

【相似文献】