Detection of oil spills using feature extraction and threshold based segmentation techniques

Satellite images can improve the possibilities for the detection of oil spills as they cover large areas and offer an economical and easier way of continuous coast areas patrolling. There are many common techniques to detect dark formations on the SAR images. This paper mainly focuses on method with spot feature extraction and global thresholding. The main approach used in this paper is detecting the dark spots, using local and global threshold algorithms. For each dark spot, a number of features are calculated in order to classify the slick as either oil or other possible geographical or natural components of water. The proposed threshold algorithm, initially analyzes the SAR images, and then assigns a probability to the dark spot to indicate whether it is an oil spill or look alike.