基于主动视觉的车辆实时检测

Real-Time Vehicle Detection Based on Active Vision

针对高速公路环境下车辆检测问题,分别对车前方的车辆和车旁超车车辆设计了基于主动视觉的车辆实时检测算法。算法首先通过标志线检测算法获得高速公路上的标志线信息,在标志线信息的引导下在低分辨率图像中通过车辆底部阴影特征搜索感兴趣区域,然后在感兴趣区域进行高分辨率图像处理,利用前方车辆的后视图灰度对称性特征和边缘特征完成前方车辆的快速检测;对于车旁超车车辆,算法在相邻车道设置检测窗口,通过检测窗口内的纹理变化信息,检测车旁超车车辆;最后利用消失点的信息进行车辆确认。实验表明,算法能快速准确地检测到公路上的车辆,具有较好鲁棒性和实时性,能够满足系统的实时性要求。

This paper presents a novel method for the detection of the front vehicle and the overtaking vehicle in highway based on active vision. At first, the algorithm detects the marking - line in highway, and locates quickly AOI （Area of Interesting） using the shadow below vehicle from the information of the marking - line in a large scale. Then, it detects fleetly the front vehicle using the symmetry and the edges of the vehicle back view in AOI in a small scale. For the overtaking vehicle, the algorithm sets a detection window on the adjacent lane, and estimates whether an overtaking vehicle exists by the information of the road surface texture. Finally, whether a vehicle appears in AOI is confirmed based on the information of vanishing point and marking line. It is shown that the approach can detect the vehicle quickly and correctly, the algorithm is robust and it meets the demand of the real time system.