Image super-resolution with PCA reduced generalized Gaussian mixture models in materials science

Single Image Super-Resolution algorithms based on patches have been noticed and widely used over the past decade. Recently, generalized Gaussian mixture models (GGMMs) shown to be a suitable tool for many image processing problems because of flexible shape parameter. In this work, we introduce supervised GGMM-based approach super-resolution two- three-dimensional images, in particular materials images. We first propose use joint GGMM learned from concatenated vectors high- low-resolution training patches. For each patch, compute minimum mean square error (MMSE) estimator generate high-resolution by averaging these estimates. select MMSE using as method is invariant affine contrast change also linear operator. Unfortunately, large dimension leads instabilities an intractable computational effort when estimating parameters GGMM. Thus, combine with principal component analysis derive EM algorithm arising model. demonstrate performance our model numerical examples synthetic real images material microstructure.