摘要
提出了一种结合多个特征的前方车辆实时检测方法.首先通过检测车底阴影特征生成车辆存在假设,然后利用车辆的垂直轮廓定位可能包含车辆的区域,最后根据车辆的灰度特征和在图像序列中出现的位置信息对感兴趣区域进行验证;使用实际采集的道路图像序列进行测试,结果表明该方法能够实时、有效地检测出前方车辆.
A preceding vehicles detecting method based on multiple features is presented.Firstly,the shadows underneath vehicles are detected to generate vehicle hypotheses.Secondly,the potential area of existing vehicle is located by the vertical outline of vehicle.Finaly,a method is proposed to judge if exist vehicles in the areas based on the gray feature of the vehicles and the location data of the vehicles in the image sequences.The method is tested with practical image sequences taken on road.The experiment results show that this method can effectively detect preceding vehicles in real time.
出处
《天津工业大学学报》
CAS
北大核心
2013年第3期72-77,共6页
Journal of Tiangong University
基金
天津市科技支撑计划重点项目(10ZCKFGX00300)
关键词
前方车辆检测
阴影特征
单目视觉
多特征
preceding vehicle detection
shadow feature
monocular vision
multiple features
