Extraction of Disaster Area from Satellite Image by combining Machine Learning and Image Processing Technology

Abstract

In recent years, heavy rain which frequently occurred in various places in Japan have been caused severe damage. It is important to identify the damaged area for disaster recovery and reconstruction. In this study, we focus on the optical satellite images that are easy to process and interpret, and extract the damaged area by combining a land cover classification method using machine learning and an additive color mixture method. As the results, it is possible to visually express the land cover changes before and after the disasters in a specific category and to extract the damaged area from the optical satellite image.

Full Paper