Abstract: The extraction of runway’s location and orientation from image sequence is required by landing navigation at night for vision-based unmanned air vehicle (UAV) and researched under universal hypothesis but without any auxiliary equipment, which reduces the dependence on the environment information and surveying instruments, enhances the independence of the system, and reduces the cost at the same time. By night, the pilot lamp of the runway can reflect the location and orientation of the runway, and these lamps which are in the same side of the runway are almost on the same line. Enough feature points can be gained after the runway image has been disposed with gray scale morphology and neighborhood processing methods. The boundary and the midline of the runway are extracted by Hough transformation. The presented method is tested with the real images, and the validity of the method is proved.

Key words: UAV, top-hat transformation, bot-hat transformation, Ostu method, Hough transformation

摘要： 在不借助外在设备和在普遍性假设的情况下，对基于视觉的无人机夜间自主着陆所需的跑道位置和方向特征提取进行了研究，降低了对外界信息和其他测量仪器的依赖性，增强了独立自主性，进一步降低了成本。对夜间机场跑道，其上的指示灯是反映其位置和方向的重要特征，这些指示灯具有明显的线性特征。通过对视频图像采用灰度形态学和邻域处理的方法进行处理，得到足够的特征点，将这些特征点所反映的跑道的边缘及中线，用Hough变换进行提取。对实际视频图像的实验证明了该方法的有效性。

关键词: 无人机, 高帽变换, 低帽变换, Ostu方法, Hough变换

CLC Number: 

• TP391