Dark Spot Detection And Characterisation Of Marine Surface Slicks Using Polsar Remote Sensing

Remote sensing using Synthetic Aperture Radar(SAR) is one of the most commonly used methods for detecting and characterising oil spills in seas and oceans. However, distinguishing true oil spills from false look-alikes like biogenic oil is a major challenge. The aim of this study is to use polarimetric decompositions like H/A/alpha decomposition and other contextual features to probabilistically separate oil spill areas from look-alikes using markov random field(MRF) classification method. The classification algorithm is applied to a qual-polarised dataset Uninhabited Ariel Vehicle Synthetic Aperture Radar(UAVSAR) of an experimental oil spill in Norway. The dataset contains a strip of plant oil and four strips of different type of oil emulsions with varying concentraion of oil. This dataset is multilooked and speckle filtered. Polarimetric features are extracted using H/A/alpha decomposition and contextual features are extracted using grey level co-occurrence matrix GLCM method. An MRF classifier is trained and its parameters are tuned using a small subset of an image. The classification results are compared before and after applying incidence angle effect correction. Using the MRF classifier on features results is a reasonable classification of different type of oil spills. The confusion between classes increases as the level of dilution in the oil-water emulsion increases. The classification accuracy improves significantly after the effect of incidence angle is corrected. Soft-classification probabilistic surface polts of oil spill area help in better characterization of oil spills. The procedure works well on classifying oil spills from water. It also helps to better characterisation of different type of oil spills.