Realtime Oil Spill Detection By Image Processing of Synthetic Aperture Radar Data

Marine transportation has high economic importance because the majority of the physical goods trades occurs via sea routes. However, minor and major oil spills can occur for a variety of natural and man-made reasons, resulting in partial or complete leaks from ships. Because of the limited opportunities for rescue in the mid-ocean, this situation has the potential to be fatal. Therefore real-time monitoring of the oil spills from ships is a huge challenge. Due to the availability of advanced satellites equipped with Synthetic Aperture Radar (SAR), it is now possible to monitor oil spills in oceans in real-time. Earlier, deep learning was widely used for this purpose; however, it requires a large amount of training data as well as computational power. We propose a novel image processing technique based on blob detection that considerably reduces computational resource usage while maintaining consistent detection accuracy. We achieved detection accuracy of up to 95.3% and detected an oil spill patch within 6.9 milliseconds. The proposed model outperforms deep learning methods by 20 times faster detection with a tradeoff of only 0.8% accuracy. The proposed modelling approach will enable automated real-time detection ship oil spills at low cost in terms of infrastructure and maintenance.