Vortex segmentation on satellite oceanographic images
Abstract
Sea temperature images coming from infrared sensors (AVHRR) allow the visualization of oceanographic activity at meso-scale (100 km), such as temperature fronts, vortices, ... These phenomena are, by their very nature, deformable, and deserve specific studies. We focused our study on the segmentation of vortices. We present three different methods to do this, using hybrid hyperquadric functions, Markovian framework, and geometrical modelization.