Automatic recognition of poleward moving auroras from all-sky image sequences based on HMM and SVM

► Automatic recognition of the poleward moving auroras from all-sky image sequences. ► We proposed a block matching-based method to estimate the auroral motion. ► The recognition is resolved based on the hybrid HMM-SVM architecture. ► Imbalanced classification problem is addressed by various metric-driven SVMs. ► Auroral observations of year 2003 at the Chinese Yellow River Station are used.