Presentation 2011-07-26
An image decomposition method by translation-invariant dictionary learning with TV regularization and its application to image inpainting
Makoto NAKASHIZUKA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, an image decomposition that represents an image as a sum of a cartoon component and texture components is proposed. TV minimization methods have been applied to the image decomposition into two components, the cartoon and the texture. We introduce the translation invariant dictionary learning with sparsity prior into the texture decomposition. The generating atoms, each of which represents a micro structure of a texture component, are learnt from the image. The texture component is decomposed into the different components with the learnt generating atoms. In experiments, the proposed image model is applied to the image inpainting, and compared with the inpainting method with a wavelet dictionary under sparsity prior.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Image decomposition / sparse signal representation / dictionary learning / total variation / image inparinting
Paper # IE2011-45
Date of Issue

Conference Information
Committee IE
Conference Date 2011/7/18(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Image Engineering (IE)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) An image decomposition method by translation-invariant dictionary learning with TV regularization and its application to image inpainting
Sub Title (in English)
Keyword(1) Image decomposition
Keyword(2) sparse signal representation
Keyword(3) dictionary learning
Keyword(4) total variation
Keyword(5) image inparinting
1st Author's Name Makoto NAKASHIZUKA
1st Author's Affiliation Graduate School of Engineering Science, Osaka University()
Date 2011-07-26
Paper # IE2011-45
Volume (vol) vol.111
Number (no) 156
Page pp.pp.-
#Pages 5
Date of Issue