基于图像处理的长距离车道线检测
2019-10-21马泉钧何自超林邦演曾文轩
马泉钧 何自超 林邦演 曾文轩



摘 要:小车在平稳路面行驶,同时前方无车辆障碍,基于此,本研究设计了长距离车道检测系统。在识别车道线时,利用输入的视频进行图像预处理,提取感兴趣区域,获取车道线的最大区域,对其进行透视变换和标定,使用Sobel算子对其二值化,使用滑块法拟合车道线并标定,输出检测后的标定视频文件。该系统能有效检测更远距离的车道线,可以应用于快速行驶路段与有清晰车道线的高速公路。
关键词:车道线检测;边缘检测;智能交通;长距离检测;透视变换
中图分类号:TP391.41文献标识码:A文章编号:1003-5168(2019)29-0111-03
Long Distance Lane Detection Based on Image Processing
MA Quanjun HE Zichao LIN Bangyan ZENG Wenxuan
(Guangzhou College of South China University of Technology,Guangzhou Guangdong 510800)
Abstract: The car is running on a smooth road with no vehicle obstacles in front. Based on this, this study designed a long-distance lane detection system. When the lane line is identified, the input video is used for image preprocessing, the region of interest is extracted, the maximum area of lane line is obtained, perspective transformation and calibration are carried out, binarization is conducted with Sobel operator, lane line is fitted and calibrated with slider method, and the calibration video file after detection is output. This system can effectively detect lane lines at longer distances and can be applied to fast-moving road sections and highways with clear lane lines.
Keywords: lane detection;edge detection;intelligent transportation;long-range detection;perspective transformation
近年来,人为因素导致车辆偏移车道而引起的交通事故屡见不鲜。中国的驾驶人整体年龄呈现上升的趋势,作为汽车主动安全技术的一部分,车道线检测对于车辆行驶安全有着重要的意义。长距离的车道线检测主要运用于高速公路等汽车行驶时速高的路段,能更早地对偏移进行检测,保证在高速行驶且制动距离较远的情况下能提前检测出偏移情况,增加驾驶员的反应时间[1-3]。车道偏移预警系统可以在形成交通事故之前对行驶情况做出检测,避免人为因素而导致车辆偏离,也可避免驾驶人疲劳而导致的交通事故。
图像预处理是车道线检测中的重要一环,其目的是更加明确地区分车道线与背景[4]。在车道线的提取中,本方案对感兴趣区域的区间进行扩大,可以处理更长的车道线。摄像头提取的图片并不是灰白图,但可以通过设置一个阈值把车道线分离出来。……
