Department of Pattern Recognition

(eng): Visual texture restoration strives not necessarily to recover the exact pixel-wise correspondence with some original unobservable texture but rather a texture which is visually indiscernible from the original one. This differs from the standard image restoration objective so it can consequently lead to different restoration techniques. A novel multispectral texture restoration method, capable to reduce simultaneously additive noise and to restore missing textural parts is presented. The restoration method is based on a descriptive, unusually complex, three-dimensional, spatial Gaussian mixture model. The model is inherently multispectral thus it does not suffer with the spectral quality compromises of the most alternative approaches.