Blind image deblurring (BID) is an ill-posed inverse problem, usuallyaddressed by imposing prior knowledge on the (unknown) image and on theblurring filter. Most of the work on BID has focused on natural images, usingimage priors based on statistical properties of generic natural images.However, in many applications, it is known that the image being recoveredbelongs to some specific class (e.g., text, face, fingerprints), and exploitingthis knowledge allows obtaining more accurate priors. In this work, we proposea method where a Gaussian mixture model (GMM) is used to learn a class-adaptedprior, by training on a dataset of clean images of that class. Experiments showthe competitiveness of the proposed method in terms of restoration quality whendealing with images containing text, faces, or fingerprints. Additionally,experiments show that the proposed method is able to handle text images at highnoise levels, outperforming state-of-the-art methods specifically designed forBID of text images.
展开▼
