RIVERINE LAND COVER CLASSIFICATION TECHNIQUE BY MACHINE LEARNING WITH SATELLITE AND UAV AERIAL IMAGES

Abstract

This paper examined a new method for classifying riverine land covers by using a machine learning technique with both the satellite and UAV(Unmanned Aerial Vehicle) images in the Kurobe River channel. The method employed Random Forests (RF) for the classification with RGBs and NDVIs(Normalized Difference Vegetation Index) of the images in combination. In the method, the high-resolution UAV images made it possible to create accurate training data for the land cover classification in the low-resolution satellite images. The results indicated that the combination of the high- and low-resolution images in the machine learning could effectively detect trees and grasses from the satellite images with a certain degree of accuracy, while the usage of only the low-resolution satellite images could not determine the difference. These results could strongly support the effectiveness of the present machine learning method for grasp the most important areas in riverine vegetation management.