A deep variational Bayesian framework for blind image deblurring

Blind image deblurring is an important but challenging problem in image processing. Traditional optimization-based methods typically formulate this task as a maximum-a-posteriori estimation or variational inference problem, whose performance highly relies on handcrafted priors for both the latent image and blur kernel. In contrast, recent deep learning methods generally learn from a large collection of training images. Deep neural networks (DNNs) directly map the blurry image to the clean image or to the blur kernel, paying less attention to the physical degradation process of the blurry image. In this study, we present a deep variational Bayesian framework for blind image deblurring. Under this framework, the posterior of the latent clean image and blur kernel can be jointly estimated in an amortized inference manner with DNNs, and the involved inference DNNs can be trained by fully considering the physical blur model, and the supervision of data driven priors for the clean image and blur kernel, which is naturally led to by the lower bound objective. Comprehensive experiments were conducted to substantiate the effectiveness of the proposed framework. The results show that it can achieve a promising performance with relatively simple networks and incorporate existing deblurring DNNs to enhance their performance.