基于序列图像的无人机进近着陆跑道识别与跟踪研究

跑道的准确识别与跟踪,是基于视觉的无人机自主着陆的重要条件.利用单目视觉获取进近着陆过程中实时跑道序列图像,经过Canny边缘检测算法和基于梯度方向的Hough变换处理,将同一跑道上的2条边缘线识别出来,并进行实时跟踪.最后利用大量真实的室外道路模拟跑道图像,以DM642为核心处理器设计视觉处理系统,结合固定翼无人机进近阶段着陆飞行特点,对设计的相关算法及视觉系统进行验证.结果表明所设计方法可以在进近着陆阶段准确提取跑道目标,并能够满足无人机着陆实时性要求.

Runway recognition and tracking study of UAV landing approach based on image sequences

The accurate recognition and tracking of the runway is the key point of the vision- based UAV autonomous landing. This paper tries to get the real- time sequence image through monocular vision,and then identify and track the two edge lines by using the canny edge detection algorithm based on the energy of the direction and Hough transform based on the gradient direction. Finally,the vision system using the DSP chip as the core processor is designed which takes advantage of a large number of real runway images based on the outdoor road to verify the relevant algorithm and the vision system,with the help of the landing approach of the fixed wing. The experiment shows that the method designed in this paper can extract the runway object accurately at the landing approach phase and satisfy the real- time request.