Supervised classification of solar features using prior information

摘要: Context: The Sun as seen by Extreme Ultraviolet (EUV) telescopes exhibits a variety of large-scale structures. Of particular interest for space-weather applications is the extraction active regions (AR) and coronal holes (CH). next generation GOESR satellites will provide continuous monitoring solar corona in six EUV bandpasses that are similar to ones provided SDO-AIA telescope since May 2010. Supervised segmentations images consistent with manual example forecasters help extracting useful information from raw data. Aims: We present supervised segmentation method based on Maximum A Posteriori rule. Our allows integrating both manually segmented well other type information. It applied segment them into AR, CH, remaining Quiet (QS) part. Methods: Bayesian classifier training masks user. noise structure nontrivial, this suggests use non-parametric kernel density estimator fit intensity distribution within each class. Under Naive Bayes assumption we can add such latitude total coverage class manner. Those be prescribed an expert or estimated Expectation-Maximization algorithm. Results: line given input show consistency over time. Introduction additional besides pixel improves upon quality final segmentation. Conclusions: Such tool aid building automated some ground truth’ defined users.