Large Scale Variational Bayesian Inference for Structured Scale Mixture Models
This work addresses image processing tasks like denoising for researchers and practitioners, but it is incremental as it builds on existing scale mixture models with a novel inference approach.
The paper tackled the problem of image denoising and inpainting by developing a large-scale variational Bayesian inference algorithm for structured scale mixture models, resulting in substantially improved performance over MAP estimation and factorial priors.
Natural image statistics exhibit hierarchical dependencies across multiple scales. Representing such prior knowledge in non-factorial latent tree models can boost performance of image denoising, inpainting, deconvolution or reconstruction substantially, beyond standard factorial "sparse" methodology. We derive a large scale approximate Bayesian inference algorithm for linear models with non-factorial (latent tree-structured) scale mixture priors. Experimental results on a range of denoising and inpainting problems demonstrate substantially improved performance compared to MAP estimation or to inference with factorial priors.