Publications
Text queries can be conducted by Author, Title, or Keyword.
Global Minimization of the Active Contour Model with
TV-Inpainting and Two-phase Denoising
Source: Lecture Notes in Computer Science 2005;3752(1):149-160.
Author: Leung SY, Osher S.
Abstract:
The active contour model [9, 10, 2] is one of the most well-known variational methods in image segmentation. In a recent paper by Bresson, Esedo¯glu, Vandergheynst, Thiran and Osher [1], a link between the active contour model and the variational denoising model of Rudin-Osher-Fatemi (ROF) [12] was demonstrated. This relation provides a method to determine the global minimizer of the active contour model. In this paper, we propose a variation of this method to determine the global minimizer of the active contour model in the case when there are missing regions in the observed image. The idea is to turn off the L1-fidelity term in some subdomains, in particular the regions for image inpainting. Minimizing this proposed energy provides a unified way to perform image denoising, image segmentation and image inpainting. To determine the minimizer of this energy functional, we use the method of gradient descent. But unlike the usual numerical method which uses the standard fully explicit scheme, we apply the Alternating Direction Explicit (ADE) scheme. This scheme provides a faster and a more robust way to minimize the proposed energy functional.
Source: Lecture Notes in Computer Science 2005;3752(1):149-160.
Author: Leung SY, Osher S.
Abstract:
The active contour model [9, 10, 2] is one of the most well-known variational methods in image segmentation. In a recent paper by Bresson, Esedo¯glu, Vandergheynst, Thiran and Osher [1], a link between the active contour model and the variational denoising model of Rudin-Osher-Fatemi (ROF) [12] was demonstrated. This relation provides a method to determine the global minimizer of the active contour model. In this paper, we propose a variation of this method to determine the global minimizer of the active contour model in the case when there are missing regions in the observed image. The idea is to turn off the L1-fidelity term in some subdomains, in particular the regions for image inpainting. Minimizing this proposed energy provides a unified way to perform image denoising, image segmentation and image inpainting. To determine the minimizer of this energy functional, we use the method of gradient descent. But unlike the usual numerical method which uses the standard fully explicit scheme, we apply the Alternating Direction Explicit (ADE) scheme. This scheme provides a faster and a more robust way to minimize the proposed energy functional.