面向近红外合作目标的鲁棒检测与匹配算法

摘要

In order to overcome the problem of cooperative object detection susceptible to disturbance in vision relative navigation for unmanned aerial vehicle (UAV )carrier landing,a robust detection and matching algorithm based on near infrared cooperative objects is proposed.Firstly,an imaging system based on near infrared cooperative objects with 8 5 0 nm is constructed.For the cooperative object image, a method of spatial geometric constraints is designed to remove the alien interference points.And Hu in-variant moments are used to further filter out the similar interference points.Finally,a matching method based on shape context is adopted to deal with the difficulty in cooperative object matching due to the dis-tortion of scale and angle.Experimental results show that the proposed algorithm can achieve cooperative object detection and matching robustly with different distances,different lights,different angles and envi-ronment disturbance.%针对无人机视觉着舰相对导航合作目标检测易受环境干扰问题，提出了一种面向近红外合作目标的鲁棒检测与匹配算法。首先，搭建了由850 nm 的近红外合作目标以及带有850 nm 滤镜摄像头组成的成像系统；然后，设计了基于几何约束的方法剔除合作目标图像中的异形干扰点，再利用 Hu 矩排除其中的相似干扰点；最后，利用形状上下文匹配算法解决拍摄图像中合作目标可能存在尺度、角度畸变使得合作目标难以匹配问题。实验结果表明，所提出的方法可以在不同距离、不同光照、不同角度以及环境干扰下，实现合作目标鲁棒检测与匹配。