摘要
针对传统车道线检测算法复杂,误检测率低,鲁棒性差等缺点,在电动试验车平台上,构建了车道线检测的硬件平台,为车辆智能驾驶系统提供可靠的车道引导信息,在对车道线检测方法分析的基础上,提出了一种能够根据实时视频流快速准确提取车道线的算法.首先对图像进行分块处理,去除掉与道路无关的区域,缩短数据处理时间,然后对图像进行灰度处理、双边滤波滤除高频噪声,再用最大类间差分(OTSU)方法得到二值图像.应用改进Hough变换检测出车道线.实际图像测试结果表明,此算法能快速准确地检测出车道线,较传统算法有更强的准确性和实时性.
As to traditional lane detection algorithm complexity,false detection rate,robustness defects,this paper builds a lane line detection hardware platform to provide reliable information on the vehicle lane guidance intelligent driving system in the electric vehicle test platform.It develops an algorithm for lane line detection with real-time video stream information based on the in-depth analysis on the technology of accurately extracting the lane.Firstly,it takes the image into blocks to get rid of the regional about unrelated road and to shorten the data processing time.The bilateral filter is used to filter out high frequency noise on the graying image.And then OTSU is introduced to obtain binary image.Improved Hough Transform is adopted to detect lanes.The practical image processing results show that the algorithm can detect the lane with more accuracy and timeliness than the traditional algorithm.
作者
王丰元
何施
徐巧妮
钟健
WANG Feng-yuan;HE Shi;XU Qiao-ni;ZHONG Jian(School of Automobile and Transportation,Qingdao University of Technology,Qingdao 266520,Chin)
出处
《青岛理工大学学报》
CAS
2018年第1期86-90,115,共6页
Journal of Qingdao University of Technology
基金
国家自然科学基金资助项目(51575288)
关键词
智能电动车
硬件平台
图像处理
车道线检测
smart electric car
hardware platform
image processing
lane line detection
