Oil Spill Detection by MODIS Images using Fuzzy Cluster and Texture Feature Extraction

Oil spills in the ocean are one of major environmental concerns, especially in the costal waters. Multispectral satellite sensors, such as AVHRR, MODIS and MERIS, have been used to detect oil spills which often exhibit a differing spectral reflectance than the surrounding waters. Some simple image processing methods, such as contrast enhancement, have been applied to remote sensing images to delineate the oil spills. But these methods often require subjective judgment from an operator and can not be used in an automatic manner, which is desirable when there is no a priori knowledge of occurrence or the spectral attributes of spills. In this study, we used a fuzzy C-means (FCM) cluster algorithm with a texture feature analysis to detect oil spill using MODIS images. The MODIS images of one incident, which happened near the new port of Dalian in Northeast China's Liaoning Province on April 3 2005, were analyzed and the results proved the efficiency of algorithm. However, in very near shore regions, delineation of water and oil boundary is noisy due to a similar spectral signature across land-water boundary.