MRI denoising based on Sparseness and Self-Similarity
Résumé
In this paper, we present two new approaches for MRI denoising. The first is an extension of the original method proposed by Guleryuz (2007). Based on local 3D DCT hard thresholding, our proposed method has been adapted to deal with Rician noise (typical of magnitude MR images) using a pseudo-oracle principle. The second proposed method is a new rotationally invariant 3D version of the Rician-adapted Non Local Means filter (Buades, 2005, Coupé 2008a, Wiest-Daesslé 2008) that uses a prefiltered image.
Domaines
Imagerie médicale| Origine | Fichiers produits par l'(les) auteur(s) |
|---|
Loading...