Learning High-Order Filters for Efficient Blind Deconvolution of Document Photographs

Lei Xiao; Jue Wang; Wolfgang Heidrich; Michael Hirsch

doi:10.1007/978-3-319-46487-9_45

Learning High-Order Filters for Efficient Blind Deconvolution of Document Photographs

Lei Xiao, Jue Wang, Wolfgang Heidrich, Michael Hirsch

Research output: Chapter in Book/Report/Conference proceeding › Chapter

37 Scopus citations

Abstract

Photographs of text documents taken by hand-held cameras can be easily degraded by camera motion during exposure. In this paper, we propose a new method for blind deconvolution of document images. Observing that document images are usually dominated by small-scale high-order structures, we propose to learn a multi-scale, interleaved cascade of shrinkage fields model, which contains a series of high-order filters to facilitate joint recovery of blur kernel and latent image. With extensive experiments, we show that our method produces high quality results and is highly efficient at the same time, making it a practical choice for deblurring high resolution text images captured by modern mobile devices. © Springer International Publishing AG 2016.

Original language	English (US)
Title of host publication	Lecture Notes in Computer Science
Publisher	Springer Nature
Pages	734-749
Number of pages	16
ISBN (Print)	9783319464862
DOIs	https://doi.org/10.1007/978-3-319-46487-9_45
State	Published - Sep 17 2016

Bibliographical note

KAUST Repository Item: Exported on 2020-10-01
Acknowledgements: This work was supported in part by Adobe and Baseline Funding of KAUST. Part of this work was done when the first author was an intern at Adobe Research. The authors thank the anonymous reviewers for helpful suggestions.

Access to Document

10.1007/978-3-319-46487-9_45

Cite this

@inbook{c9678442ce3a4912a8b9cd93c5b6b073,

title = "Learning High-Order Filters for Efficient Blind Deconvolution of Document Photographs",

abstract = "Photographs of text documents taken by hand-held cameras can be easily degraded by camera motion during exposure. In this paper, we propose a new method for blind deconvolution of document images. Observing that document images are usually dominated by small-scale high-order structures, we propose to learn a multi-scale, interleaved cascade of shrinkage fields model, which contains a series of high-order filters to facilitate joint recovery of blur kernel and latent image. With extensive experiments, we show that our method produces high quality results and is highly efficient at the same time, making it a practical choice for deblurring high resolution text images captured by modern mobile devices. {\textcopyright} Springer International Publishing AG 2016.",

author = "Lei Xiao and Jue Wang and Wolfgang Heidrich and Michael Hirsch",

note = "KAUST Repository Item: Exported on 2020-10-01 Acknowledgements: This work was supported in part by Adobe and Baseline Funding of KAUST. Part of this work was done when the first author was an intern at Adobe Research. The authors thank the anonymous reviewers for helpful suggestions.",

year = "2016",

month = sep,

day = "17",

doi = "10.1007/978-3-319-46487-9_45",

language = "English (US)",

isbn = "9783319464862",

pages = "734--749",

booktitle = "Lecture Notes in Computer Science",

publisher = "Springer Nature",

}

TY - CHAP

T1 - Learning High-Order Filters for Efficient Blind Deconvolution of Document Photographs

AU - Xiao, Lei

AU - Wang, Jue

AU - Heidrich, Wolfgang

AU - Hirsch, Michael

N1 - KAUST Repository Item: Exported on 2020-10-01 Acknowledgements: This work was supported in part by Adobe and Baseline Funding of KAUST. Part of this work was done when the first author was an intern at Adobe Research. The authors thank the anonymous reviewers for helpful suggestions.

PY - 2016/9/17

Y1 - 2016/9/17

N2 - Photographs of text documents taken by hand-held cameras can be easily degraded by camera motion during exposure. In this paper, we propose a new method for blind deconvolution of document images. Observing that document images are usually dominated by small-scale high-order structures, we propose to learn a multi-scale, interleaved cascade of shrinkage fields model, which contains a series of high-order filters to facilitate joint recovery of blur kernel and latent image. With extensive experiments, we show that our method produces high quality results and is highly efficient at the same time, making it a practical choice for deblurring high resolution text images captured by modern mobile devices. © Springer International Publishing AG 2016.

AB - Photographs of text documents taken by hand-held cameras can be easily degraded by camera motion during exposure. In this paper, we propose a new method for blind deconvolution of document images. Observing that document images are usually dominated by small-scale high-order structures, we propose to learn a multi-scale, interleaved cascade of shrinkage fields model, which contains a series of high-order filters to facilitate joint recovery of blur kernel and latent image. With extensive experiments, we show that our method produces high quality results and is highly efficient at the same time, making it a practical choice for deblurring high resolution text images captured by modern mobile devices. © Springer International Publishing AG 2016.

UR - http://hdl.handle.net/10754/622213

UR - http://link.springer.com/chapter/10.1007%2F978-3-319-46487-9_45

UR - http://www.scopus.com/inward/record.url?scp=84990048809&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-46487-9_45

DO - 10.1007/978-3-319-46487-9_45

M3 - Chapter

SN - 9783319464862

SP - 734

EP - 749

BT - Lecture Notes in Computer Science

PB - Springer Nature

ER -

Learning High-Order Filters for Efficient Blind Deconvolution of Document Photographs

Abstract

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this