Learning Blind Deconvolution

In this work, we propose a novel prior term for the regularization of blind deblurring methods. The proposed method introduces machine learning techniques into the blind deconvolution process. The proposed technique has sound mathematical foundations and is generic to many inverse problems. We demonstrate the usage of this regularizer within Bayesian blind deconvolution framework, and also integrate into the latter a method for noise reduction, which was previously proposed in the context of non-blind astronomical image deblurring, thus creating a complete blind deconvolution method. The application of the proposed algorithm is demonstrated on three-dimensional images acquired by a wide-field fluorescence microscope, where the need for blind deconvolution algorithms is indispensable, yielding excellent results.