Automatic detection of oil spills by SAR images : dark spot detection and feature extraction

Abstract

This report is written in the context of a PhD-study. The research is aiming at exploiting the potential of automatic detection of oil spills using new, advanced algorithms to demonstrate how well such a system will perform in terms of accuracy on a large set of SAR images. The approach to automatic oil spill detection has three main parts: dark spot detection, feature extraction and spot classification. This report covers some of the most recent contributions to the dark spot detector and the feature extractor.