摘要
为帮助驾驶员发现目标车道侧后方车辆尤其是视野盲区内车辆,从而避免或减少换道过程中发生的交通事故,提出了日间在高速公路上识别侧后方车辆的新方法.根据路面灰度值,将日间图像分为正常光照和弱光照两类.由于车辆阴影的灰度小于路面平均灰度,根据两侧车道区域内存在的灰度突变确定出侧后方车辆的可能存在区域.对确定的区域采用相应的阈值分割方法进行图像分割.在对分割后的二值化图像去噪、边缘提取和细化,以及提取车辆直线水平边缘基础上,根据系列车辆统计获得的先验知识(车辆前车窗的大小及比例等)验证车辆的真实存在.试验结果表明,该算法具有较好的可靠性和鲁棒性.
A method for detection of backside vehicle on freeway in daytime was proposed to help the driver check the existence of backside vehicles in the target lane,especially those in the blind zones,and to reduce traffic accidents during lane changing.The road images are classified into two categories： taken under normal and weak illumination conditions according to their gray values.The possible backside vehicle is recognized if there is a sudden change in gray values within the area in the lanes of both sides because the gray values of vehicle shadows are smaller than the average gray value of the road surface.The existence of backside vehicle is confirmed by comparison between the ratios of windshield sizes extracted from the images with those in a list of the known vehicles after segmentations with different thresholds for normal and weak illumination conditions,denoising,and boundary extraction.Experimental results show that the proposed algorithm has good reliability and robustness.
出处
《西南交通大学学报》
EI
CSCD
北大核心
2010年第2期231-237,共7页
Journal of Southwest Jiaotong University
基金
国家自然科学基金资助项目(50908098)
吉林省国际科技合作项目(20080706)
吉林大学基本科研业务专项资金项目的支持
关键词
交通运输
安全工程
盲区
车辆识别
灰度
transportation
safety engineering
blind zone
vehicle detection
gray value
