Study on the Automatic Recognition of Oceanic Eddies in Satellite Images by Ellipse Center Detection—The Iberian Coast Case

Ellipse detection has been presented several times in scientific literature as a potential method for finding eddies in satellite images because eddies may be roughly approximated by ellipses. In this paper, therefore, we describe a new eddy detector based on ellipse detection. The detector is capable of finding several eddies per satellite image, an ability not demonstrated for previously reported eddy detectors based on ellipse detection in the morphology of binary images. One important characteristic of this paper is the determination of the method's classification efficiency using a considerable number of eddies; such a determination has been rarely performed in the scientific literature on eddy detection. The developed detector finds ellipse centers using geometrical properties of ellipses. To the best of the author's knowledge, the proposed method has never been applied to eddy detection. The low value of the maximum temperature gradient of the eddies off the Iberian Peninsula (0.52degC/pixel) complicates the application of the detection method because it raises difficulties in outlining the eddies and induces large amounts of noise in the binary images from which the eddies are detected. The values of each parameter in the developed system were equal for the images analyzed, which suggests that the application of the proposed system to large numbers of images is simple. The author did not find any method applicable to the present case in the scientific literature whose classification efficiencies, assessed on complete images or a considerable number of eddies, were better than those reported here.