Small UAV Based Multi-Viewpoint Image Registration for Extracting the Information of Cultivated Land in the Hills and Mountains

Abstract

The amount of arable land in southern China is reduced due to land degradation and soil erosion. Arable land change by remote sensing technology is the most economical and efficient way to relieve the pressure of agricultural production. Therefore, we present a small unmanned aerial vehicle (U A V) based multi-viewpoint image registration method for extracting the information of arable changes in hills and mountains. Three major contributions of our method are included: (i) feature point sets were extracted by SURF; (ii) reliable correspondence was established by mixture-feature finite mixture model (MFMM); (iii) $Lz$-minimizing estimate $(L_{2}E)$ based energy function with double geometric constraints was used to estimate the transformation function. Compared with five state-of-the-art methods, our method shows better performances in most cases.