Detection of oil spills in a complex scene of SAR imagery

We present a method for detecting oil spills in a complex scene of SAR imagery, including segmenting oil spills, and avoiding false alarms. Segmentation is carried out using a multi-time and multi-hierarchical method by dividing the complex sea surface into bright sea and dark sea. Gray-based and edge-based segmentations are done to extract oil spills from bright and dark sea, respectively. The proposed method can extract complete oil spills, obtain better visual results, and increase detection probability more accurately than the traditional method. Based on the surrounding features and the oil spills' features, dark land spots and low contrast dark spots are removed efficiently, thus reducing false alarms. The experimental results demonstrate that the proposed algorithm has fast computation speed, high detection accuracy, and is very useful and effective for detecting oil spills in SAR imagery.