摘要
利用最大方差阈值方法对灰度图像二值化后,对车辆底部阴影突变行检测,对感兴趣区域确定方法进行改进,实现了感兴趣区域的精确确定.通过计算精确确定后的感兴趣区域的灰度对称性和熵值归一化对称性测度,实现了车辆的精确检测,找到车辆的对称轴,并在车辆区域上边界一定范围内,进行灰度突变检测,实现了车辆上边界的准确定位.通过实验验证算法较好地实现了复杂环境下的车辆检测和上边界定位.
The information of surrounding was gotten by the way of vision technology in intelligent vehicle system.It is very important to detcect other vehicles in front of the ego vehicle on road.The method of vehicle detection was proposed.The method of Ostu threshold was used to realize image binarization.Searching line may be the line of the shadow of vehicle＇s bottom as the AOI＇s（Area of Interest） bottom boundary.The expanding of the AOI was realized.The symmetry measurement and symmetry measurement of normalized entropy were used to determine whether the region is vehicle or not.The upper boundary was determined by checking the gray mutant line in the gray image.This method was tested,and got good results.The method paves the way for the latter distance measurement between vehicles.
出处
《北京航空航天大学学报》
EI
CAS
CSCD
北大核心
2010年第7期781-784,共4页
Journal of Beijing University of Aeronautics and Astronautics
基金
教育部高等学校博士学科点专项科研基金资助项目(20070006011)
关键词
计算机视觉
底部阴影
车辆检测
computer vision
vehicular shadow information
vehicle detection
