Texture Analysis for Oil Spill Segmentation Based on X-band Marine Radar Imagery

We propose an index of textural feature for oil spill segmentation based on X-band marine radar image. The radar image used in this paper were acquired on July 21, 2010, from the teaching-training ship, "YUKUN", of Dalian Maritime University. Using adaptive median filter and Fields-of-Experts model, co-channel interference and small objects are cleaned separately. An index calculated by textural features by grey level co-occurrence matrix, including entropy, energy, correlation and contrast, is used to indicate the region containing oil slicks. With the proposed index of textural features, oil spills extracted by texture analysis is comparable to visual interpretation.