How Diffusion Prior Landscapes Shape the Posterior in Blind Deconvolution
This addresses a fundamental limitation in blind deconvolution for image restoration, offering insights for improved priors and optimization techniques, though it appears incremental as it builds on existing MAP framework analysis.
The paper tackles the problem of blind deconvolution in image processing, where Maximum A Posteriori (MAP) estimation with sparsity-promoting priors tends to favor blurry solutions. By analyzing diffusion-based priors, they show that local minimizers of the posterior correspond to realistic sharp images, effectively solving the problem.
The Maximum A Posteriori (MAP) estimation is a widely used framework in blind deconvolution to recover sharp images from blurred observations. The estimated image and blur filter are defined as the maximizer of the posterior distribution. However, when paired with sparsity-promoting image priors, MAP estimation has been shown to favors blurry solutions, limiting its effectiveness. In this paper, we revisit this result using diffusion-based priors, a class of models that capture realistic image distributions. Through an empirical examination of the prior's likelihood landscape, we uncover two key properties: first, blurry images tend to have higher likelihoods; second, the landscape contains numerous local minimizers that correspond to natural images. Building on these insights, we provide a theoretical analysis of the blind deblurring posterior. This reveals that the MAP estimator tends to produce sharp filters (close to the Dirac delta function) and blurry solutions. However local minimizers of the posterior, which can be obtained with gradient descent, correspond to realistic, natural images, effectively solving the blind deconvolution problem. Our findings suggest that overcoming MAP's limitations requires good local initialization to local minima in the posterior landscape. We validate our analysis with numerical experiments, demonstrating the practical implications of our insights for designing improved priors and optimization techniques.