A Super Resolution Algorithm for Atmospherically Degraded Images Using Lucky Regions and MAP-uHMT

This paper demonstrates the possibility of super resolved image reconstruction for images affected by atmospheric turbulence. A lucky region method using bicoherence is proposed to select image tiles with superior quality or ¿lucky image regions¿ from a large number of short exposure images. A super resolved image is then reconstructed by a MAP method based on a universal hidden Markov tree model from the lucky regions. Performance is demonstrated with real data.