Abstract
A new method designed for the detection of mesoscale structures in sea surface temperature (SST) satellite images, to be used in different applications such as climatic and environmental studies or fisheries, is presented. The method is based on an entropic approach technique to edge detection, using the Jensen–Shannon divergence. It is found to be an excellent edge detector technique that exhibits favorable characteristics. For example, it is very robust against impulsive and Gaussian noise, avoiding the use of previous filtering, with the subsequent gain in computational work and edge sharpness. The method is evaluated on a set of Advanced Very High-Resolution radiometer images of the Atlantic Ocean and Mediterranean Sea, near the Iberian Peninsula area. The SST fields have been generated using a split-window technique to avoid the problem of atmospheric disturbance; emissivity correction has been carried out to improve the reliability of the data. The results have been compared to those obtained using several methods proposed in the literature. Some of the images were corrupted with impulsive noise before the processing to show the robustness of the method.
Corresponding author address: Dr. David Pozo Vázquez, Departamento de Física Aplicada, Universidad de Granada, Facultad de Ciencias, 18071, Avda de Fuente Nueva, Granada, Spain.
Email: dpozo@golat.ugr.es
