A NOVEL THRESHOLD BASED APPROACH OF DETECTING OIL SPILLS ON SEA IN SYNTHETIC APERTURE RADAR IMAGES

Abstract

Nowadays, phenomena of oil spills commonly take place on rivers and sea, and cause severe consequences on the water environment. Therefore, detecting oil spills and providing early warnings of them have received great interests for recent decades. There have been many algorithms developed to identify oil spills using the Synthetic Aperture Radar images because of their quality independence of weather conditions and capability of event capture of a wide geometry range. Among them, the threshold based methods are quite popular in reality because of their implementation simplicity. However, these algorithms provide relatively low accuracy. The paper proposed a novel threshold based algorithm of oil spill detection in the Synthetic Aperture Radar images. This threshold is a global one determined according to statistical analysis of pixel intensities of the images and their sizes. The simulation results of the proposed method on Python software were compared with other methods, and proved that the proposed method significantly improved accuracy.