摘要
为了提高无人驾驶汽车视觉导航系统中车道线检测的准确性和实时性,在对车道线检测技术进行深入研究的基础上,提出一种能快速准确检测出车道线的新算法。首先采用分块思想将RGB图像中与道路无关的区域去除,以缩短数据处理时间。然后对余下的RGB图像进行灰度化处理,接着用中值滤波法消除随机噪声,再用最大类间方差法(Otsu法)初步得到二值图像。最后对二值图像利用数学形态学进一步边缘细化,使位于车道线上的每个像素行只有一个像素特征点,再采用Hough变换检测出车道线。Matlab仿真结果表明,此算法能够快速准确地检测出车道线,较传统检测算法具有更强准确性和实时性。
In order to improve accuracy and real-time of lane line detection in vision navigation system of unmanned vehicle,based on an in-depth study of the lane line detection technique,a new algorithm for fast and accurate detection of lane is introduced. Firsly,through the use of block theory,the road independent region of RGB image is removed,to shorten the data processing time. And then after the graying of the rest of RGB image,the median filtering method is used to eliminate the random noise,the Method of Maximum Classes Square Error（Method of Otsu）is introduced to obtain the binary image. Finally,edge thinning based on mathematical morphology is proposed to make each pixel row of each lane only left one characteristic pixel. And the lane line detection based on Hough transformation is carried out. The Matlab simulation results show that the proposed algorithm can detect lane line quickly and accurately,compared with the traditional method has better accuracy and real-time.
出处
《火力与指挥控制》
CSCD
北大核心
2015年第6期152-154,158,共4页
Fire Control & Command Control
基金
山西省安监局基金资助项目
关键词
无人驾驶
阈值分割
边缘检测
数学形态学
HOUGH变换
unmanned vehicle
threshold segmentation
edge detection
mathematical morphology
hough transformation
