一种改进的图像拼接技术及其在无人机病虫害检测中的应用

An Improved Image Mosaic Technique and Its Application in UAV Pest and Disease Detection

为解决ORB(oriented FAST and rotated BRIEF)算法提取的特征点分布不均匀且可能重叠的问题,在ORB算法的基础上引入了四叉树结构实现特征点均匀分配,使用KNN算法进行特征点粗略匹配,通过RANSAC算法去除虚假匹配点以获得完整的病虫害图像.实验结果表明,与ORB算法和SIFT相比,新算法匹配精度分别提高了8%以上和3%以上.

In order to solve the problem of uneven distribution of feature points extracted by ORB(oriented FAST and rotated BRIEF)algorithm and possible overlap,a quadtree structure is introduced on the basis of ORB algorithm to achieve uniform distribution of feature points,KNN algorithm is used for rough matching of feature points,and false matching points are removed by RANSAC algorithm to obtain a complete pest and disease images.The experimental results show that this paper improves the matching accuracy by more than 8%and 3%compared with the ORB algorithm and SIFT,respectively.