Current Issue Cover
基于单目视觉的实时测距方法研究

郭磊1, 徐友春1, 李克强1, 连小珉1(清华大学汽车安全与节能国家重点实验室,北京 100084)

摘 要
为了利用单目视觉实时监测本车与前方障碍物之间的距离,在比较了现有的几类用于车辆控制的道路深度信息获取方法的基础上,首先研究了较为适用于汽车自动驾驶的几何关系推导法,进而提出了基于单目视觉的实时测距算法.通过试验可知,由于摄像机的俯仰角是影响实时测距算法的关键因素,因此又提出了基于道路边界平行约束条件的实时计算摄像机俯仰角算法.静态实车试验的结果显示,该基于单目视觉的实时测距算法具有较高的准确性,可以满足测距要求,而动态实车试验的结果则显示,此算法还可以满足汽车智能化控制的实时性要求.
关键词
Study on Real-time Distance Detection Based on Monocular Vision Technique

()

Abstract
This paper dedicates to develop an algorithm which can be used to detect the real-time distance between the own car and the obstacle ahead. On the basis of comparing several existing methods, this paper uses the geometrical reasoning method to obtain the depth information of the driving environment for the vehicle control, and presents a real-tlme monocular-vision-based algorithm to get the distance between the own car and the obstacle ahead. The pitch angle of the camera is the key factor which affects the precision of the distance detection. This conclusion is drawn from the experimental results. In order to get the real pitch angle of the camera, the constraint that the two lane boundaries are parallel is used in the algorithm. The results of the static real-car tests show that the algorithm presented in this paper satisfies the precision requirements of the distance detection. And the results of the dynamic real-car tests show that the algorithm also satisfies the real-time requirements for the intelligent vehicle control.
Keywords

订阅号|日报