Methodology for eddies recognition from satellite images

In this this work is presented a semi-automatically technique for eddies recognition using GOES Sea Surface Temperature (SST) data time series from 2003 to 2008. The methodology consists of the main steps as follows: 1) identify the main characteristics of the eddies and define the area of interest (such climatology was derived from SST images for the period of 2003 to 2008); 2) cluster SST images; and 3) develop a morphological classifier based on a structuring element for identification of eddies. As a part of pre-processing, the fuzzy c-means clustering has demonstrated good results, followed by a morphology classification. The results are quite good ones and have demonstrated that is possible to develop and implement an efficient classifier.