In the recent years, the power of employing singularity analysis of scalar variables to perform pattern recognition in complex images has been evidenced. This approach is particularly useful when the image under study corresponds to a scalar variable submitted to a turbulent flow because, in that case, the arrangement of singularity manifolds corresponds to the multifractal hierarchy from the underlying turbulent flow. In this paper we will show how this intimate connection between Image Processing and Physics, when applied to Microwave Sea Surface Temperature images, allows to uncover global circulation patterns in the ocean at a daily basis with the resolution of 1/4 of degree. Using this technique, details not evidenced in the non-processed image are revealed. The emerging patterns provide a description of the global oceanic currents much richer than the usual global circulation scheme; in particular, instabilities, eddy generation and filamentation are distinctly identified. A pattern extraction of this kind is useful in order to measure and track mesoscale oceanic phenomena, opening the way to many operational and reanalysis applications.