Non-rigid Image Feature Matching for Unmanned Aerial Vehicle in Precision Agriculture

Abstract

In this paper, we propose a novel feature matching method for agricultural UAV based on probabilistic inference when they undergo non-rigid transformations. Firstly, a set of putative correspondences was generated based on the similarity of features. Then, we focus on eliminating outliers from that set meanwhile estimating the transformation. This procedure is formulated upon a maximum likelihood Bayesian model. We impose three effective regularization techniques on the correspondence, which helps to find an optimal solution. The problem is finally addressed rely on the EM algorithm. Extensive experiments on real farm images shows accurate results of our method, which is superior to the current state-of-the-art methods.