Unsupervised segmentation of low clouds from infrared METEOSAT images based on a contextual spatio-temporal labeling approach

The early and accurate segmentation of low clouds during the night-time is an important task for nowcasting. It requires that observations can be acquired at a sufficient time rate as provided by the geostationary METEOSAT satellite over Europe. However, the information supplied by the single infrared METEOSAT channel available by night is not sufficient to discriminate between low clouds and ground during night from a single image. To tackle this issue, we consider several sources of information extracted from an infrared image sequence. Indeed, we exploit both relevant local motion-based measurements, intensity images and thermal parameters estimated over blocks, along with local contextual information. A statistical contextual labeling process in two classes, involving “low clouds” and “clear sky,” is performed on the warmer pixels. It is formulated within a Bayesian estimation framework associated with Markov random field (MRF) models. This comes to minimize a global energy function comprising three terms: two data-driven terms (thermal and motion-based ones) and a regularization term expressing a priori knowledge on the label field (expected spatial contextual properties). We propose a progressive minimization procedure of this energy function starting from initial reliably labeled pixels and involving only local computation. Thermal parameters associated to each class are estimated according to an unsupervised learning scheme enabling the handling of spatiotemporal nonstationarities. Our method produces segmentation maps displaying temporal coherency along the image sequence. Experimental results on representative meteorological situations are reported and favorably compared with NOAA/AVHRR cloud classifications which serve as reference results. They demonstrate the accuracy and efficiency of the proposed approach.