Detecting Dark Spot From Oil Spill SAR Intensity Images By Point Process With Irregular Geometry Mark

Oil spill has become the most important pollutant on the sea surface.In Synthetic Aperture Radar(SAR)image the sea surface areas covered by oil spill appear dark spots.Therefore,the main task of the oil spill detection in SAR image is to extract the dark spots,which lays the foundation for the subsequent oil spill identification.Due to the characteristics of continual drift and diffusion over time,it is difficult to determine the geometric shape of oil spill,this also causes the dark spot extraction becomes more difficult.Aiming at the difficulty to determine the geometric shape of the dark spot,this paper thesis proposes a method of dark spot extraction from SAR images based on marked point process theory.Firstly,a set of random points is defined in SAR image domain used to describe the locations of dark spots,and an irregular polygon is associated for each random point as the shape of the dark spot indicated by the point.Secondly,assume that intensities of pixels in and out of different irregular polygons obey their own independence and identity Gamma distributions,respectively.Then the statistical distribution model for the intensities SAR image is built.Thirdly,by combining the two above models,the dark spot extraction model is constructed based on Bayesian theory.Finally,Reversible Jump Markov Chain Monte Carlo(RJMCMC)algorithm is designed to simulate the dark spot extraction model,and then the optimal solution for the locations,distributions,shapes and statistic model parameters of the dark spots is obtained according to Maximum A Posteriori(MAP)scheme.To verify the validness of the proposed method,testing is carried out with simulated and real SAR images.The results demonstrate that the proposed algorithm can not only locate and extract the dark spots,but also accurately estimate the statistical distribution parameters of each dark spot region.