I hacked away on the blind-deconvolution formulation of Lucky Imaging. I have reimplemented the online blind deconvolution method of Hirsch et al but added regularization and also taking multiple passes through the data (thus getting a better solution but breaking the online-iness). The code is now a very elegant core surrounded by lots of hacky, heuristic code. Not happy about that, but it gets the job done.
No comments:
Post a Comment