Data-driven image completion by image patch subspaces

Author

Mobahi, Hossein ; Rao, Shankar R. ; Ma, Yi

Author_Institution

Coordinated Sci. Lab., Univ. of Illinois at Urbana Champaign, Champaign, IL, USA

fYear

2009

fDate

6-8 May 2009

Firstpage

1

Lastpage

4

Abstract

We develop a new method for image completion on images with large missing regions. We assume that similar patches form low dimensional clusters in the image space where each cluster can be approximated by a (degenerate) Gaussian. We use sparse representation for subspace detection and then compute the most probable completion. Our results show almost no blurring or blocking effects. In addition, both the texture and structure of the missing regions look realistic to the human eye.

Keywords

Gaussian distribution; Gaussian processes; approximation theory; image reconstruction; image texture; pattern clustering; signal detection; Gaussian distribution; approximation theory; data-driven image completion; image inpainting; image patch subspace; image texture; large missing region; low dimensional cluster; sparse representation; subspace detection; Dictionaries; Eyes; Filling; Humans; Image reconstruction; Image restoration; Mathematical model; Partial differential equations; Training data; Transforms; Degenerate Gaussians; Image Subspaces; Inpainting; Sparse Representation;

fLanguage

English

Publisher

ieee

Conference_Titel

Picture Coding Symposium, 2009. PCS 2009

Conference_Location

Chicago, IL

Print_ISBN

978-1-4244-4593-6

Electronic_ISBN

978-1-4244-4594-3

Type

conf

DOI

10.1109/PCS.2009.5167452

Filename

5167452