摘要
提出一种融合多种目标特征的单目视觉车辆检测方法。首先,利用车辆尾部的结构对称性提取出感兴趣区域(ROI),减少搜索范围;然后,利用车辆底部的阴影特征,在ROI中搜寻车辆可能出现的位置,找出假设目标;最后,利用亮度和轮廓信息对假设目标进行对称性验证,排除虚假目标,同时对车辆在图像中的位置实现精确定位。通过实验,验证了提出方法的有效性和鲁棒性。
A novel on-road vehicle detection approach with monocular camera is presented which fuses multi-cues of object.First,the horizontal symmtery of vehicle rear view is utilized to achieve the region of interest so as to reduce the search area of following process.And then,the sign of underneath shadow is employed to generate hypothetical positions on which potential vehicles maybe present.Finally,both image intensity and figure information are combined to verify the vertical symmetry of the potential vehicle candidates.As a result,the vehicle can be localizated and some false alarms will be rejected.Experimental results validate that the proposed approach is robust and effective under real-world video data.
出处
《光电子．激光》
EI
CAS
CSCD
北大核心
2010年第1期74-77,共4页
Journal of Optoelectronics·Laser
基金
国家自然科学基金资助项目(50675099)
江苏省自然科学基金资助项目(BK2007197)
江苏省普通高校研究生科研创新计划资助项目(CX08B-044Z)
关键词
机器视觉
车辆检测
智能车辆
智能交通系统
对称性
machine vision
vehicle detection
intelligent vehicles
intelligent traffic system
symmetry
