Oil Slick Identification in Marine Radar Image Using HOG, Random Forest and PSO

Marine oil spills have become a significant threat to ocean environments and ecosystems. The development of effective marine oil spill identification technology is crucial for emergency authorities to enhance response strategies. In this paper, a marine oil spill detection method based on Histogram of Oriented Gradient (HOG) features, random forest classifier and particle swarm optimization (PSO) algorithm was proposed. Firstly, row vector convolution and mean filtering were adopted to extract and smooth the co-frequency interferences. Then, binarization processing and median filtering were used to eliminate speckles. After that, gray correction and contrast enhancement model were performed to enhance the overall oil slick features. Next, HOG features and Random Forest classifier were utilized to extract effective oil spill regions. Finally, the PSO algorithm was introduced to iteratively optimize the adaptive dual-threshold for segmenting the real oil slicks. This approach enables accurate and efficient detection of oil spills, offering a scientific foundation for responding to offshore oil spill incidents.