Ship detection using optical satellite imagery

Abstract

Given Singapore is a coastal state that connects the West and the East. Being one of the world’s busiest port, maritime security plays a vital role. A comprehensive maritime monitoring system should be developed so as to accommodate growing traffic demand. This project aims at automated extraction of sea-going vessels using the optical imaging products of 21AT’s TripleSat Constellation. Traditionally, synthetic-aperture radar (SAR) satellite imagery is used due to its being “All Weather, All Day” and its prominent target reflections that make ship detection fast and robust. However, optical satellite imagery has its edge over the SAR counterpart in the choices of bands, resolutions and revisit rates, let alone the costs. The implementation of this project leverage on Matlab with image processing toolbox. There are two filtering conditions involved in this Matlab program, which are length-beam ratio and roundness. By using these two constrains, the program can successfully detect sea-going ships among all those false alarms like small island, wakes and cloud. However, due to the limitations of this research, other factors are not considered. For future research, it is highly recommended to have more filtering criteria and apply machine learning in the algorithm.