Image Denoising and Registration by PDE's on the Space of Patches.

David Tschumperlé (1) , Luc Brun (1)

1 Equipe Image - Laboratoire GREYC - UMR6072

We define a simple transform able to map any multi-valued image into a space of patches, such that each existing image patch is mapped into a single high-dimensional point. We show that solving variational problems on this particular space is an elegant way of finding the natural patch-based counterparts of classical image processing techniques, such as the \emph{Tikhonov} regularization and \emph{Lucas-Kanade} registration methods. We end up with interesting variants of already known (non-variational) patch-based algorithms, namely the \emph{Non Local Means} and \emph{Block Matching} techniques. The interest of considering variational approaches on patch spaces is discussed and illustrated by comparison results with corresponding non-variational and non-patch methods.

Domaines: Traitement des images [eess.IV]

David Tschumperlé, Luc Brun. Image Denoising and Registration by PDE's on the Space of Patches.. International Workshop on Local and Non-Local Approximation in Image Processing (LNLA'08), 2008, Lausanne, Switzerland. pp.32--40. ⟨hal-00332804⟩