From: Bjørn Rustad Date: Mon, 15 Sep 2014 07:44:33 +0000 (+0200) Subject: Oppgavetekst til kontrakten X-Git-Url: http://git.rustad.me/?a=commitdiff_plain;h=935fff35a5806f98d253f4e08a75ae62c7562ac7;p=master Oppgavetekst til kontrakten --- diff --git a/oppgavetekst b/oppgavetekst new file mode 100644 index 0000000..e00f1ba --- /dev/null +++ b/oppgavetekst @@ -0,0 +1,18 @@ +Anisotropic total variation based image restoration using graph cuts + +Anisotropic diffusion and total variation are two popular methods in +image restoration, where the goal is to remove noise from a captured +image. They are both popular because of their ability to preserve edges +in the original image, but also have certain problems. Anisotropic +diffusion filtering may introduce new structures from the noise, while +total variation filtering may reduce the contrast near corners and thin +structures. The objective of the thesis is to introduce an anisotropic +total variation norm enforcing strong regularization along edges and +weak regularization across edges. The resulting energy function will +then be minimized using a graph cut approach based on the one that is +popular in regular total variation filtering. In the anisotropic +setting, the necessary edge weights can be constructed by using an +anisotropic variant of the Cauchy Crofton formula. The resulting method +is then be compared to regular anisotropic diffusion and total variation +filtering. +