Ship Detection from Sentinel-1 Imagery through Thresholding and Clustering Method

Abstract

Maritime surveillance has been a significant issue for protecting illegal activities at sea and conserving marine resources. However, the Automatic Identification System and V-Pass, a fishing boats' position reporting system, have a limited transmission distance and a high dependency on volunteering message reporting. Therefore, satellite imagery-based ship detection has emerged in maritime surveillance in recent decades. This study proposes a ship detection approach through the combination method of thresholding and clustering (TCM). Sentinel-1 imageries were used, that freely available at and provided by the Copernicus Open Access Hub that operated by European Space Agency (ESA). Ships at sea are detected through TCM after completing preprocessing procedures, including thermal noise removal, terrain correction, and masking out of a land area. The proposed method shows a high detection speed and is expected to contribute to maritime surveillance.