skip to main content
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Variational Dirichlet Blur Kernel Estimation

Abstract

Blind image deconvolution involves two key objectives: 1) latent image and 2) blur estimation. For latent image estimation, we propose a fast deconvolution algorithm, which uses an image prior of nondimensional Gaussianity measure to enforce sparsity and an undetermined boundary condition methodology to reduce boundary artifacts. For blur estimation, a linear inverse problem with normalization and nonnegative constraints must be solved. However, the normalization constraint is ignored in many blind image deblurring methods, mainly because it makes the problem less tractable. In this paper, we show that the normalization constraint can be very naturally incorporated into the estimation process by using a Dirichlet distribution to approximate the posterior distribution of the blur. Making use of variational Dirichlet approximation, we provide a blur posterior approximation that considers the uncertainty of the estimate and removes noise in the estimated kernel. Furthermore, experiments with synthetic and real data demonstrate that the proposed method is very competitive to the state-of-the-art blind image restoration methods.

Authors:
 [1];  [2];  [1];  [2]; ORCiD logo [3]
  1. Beihang Univ., Beijing (China)
  2. Univ. de Granada, Granada (Spain)
  3. Northwestern Univ., Evanston, IL (United States)
Publication Date:
Research Org.:
Northwestern Univ., Evanston, IL (United States). Center for Catalysis and Surface Science
Sponsoring Org.:
USDOE National Nuclear Security Administration (NNSA), Office of Nonproliferation and Verification Research and Development (NA-22)
OSTI Identifier:
1487984
Grant/Contract Number:  
NA0002520
Resource Type:
Accepted Manuscript
Journal Name:
IEEE Transactions on Image Processing
Additional Journal Information:
Journal Volume: 24; Journal Issue: 12; Journal ID: ISSN 1057-7149
Publisher:
IEEE
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS AND COMPUTING; blind deconvolution; image deblurring; variational distribution approximations; Dirichlet distribution; constrained optimization; point spread function; inverse problem

Citation Formats

Zhou, Xu, Mateos, Javier, Zhou, Fugen, Molina, Rafael, and Katsaggelos, Aggelos K. Variational Dirichlet Blur Kernel Estimation. United States: N. p., 2015. Web. doi:10.1109/TIP.2015.2478407.
Zhou, Xu, Mateos, Javier, Zhou, Fugen, Molina, Rafael, & Katsaggelos, Aggelos K. Variational Dirichlet Blur Kernel Estimation. United States. doi:10.1109/TIP.2015.2478407.
Zhou, Xu, Mateos, Javier, Zhou, Fugen, Molina, Rafael, and Katsaggelos, Aggelos K. Mon . "Variational Dirichlet Blur Kernel Estimation". United States. doi:10.1109/TIP.2015.2478407. https://www.osti.gov/servlets/purl/1487984.
@article{osti_1487984,
title = {Variational Dirichlet Blur Kernel Estimation},
author = {Zhou, Xu and Mateos, Javier and Zhou, Fugen and Molina, Rafael and Katsaggelos, Aggelos K.},
abstractNote = {Blind image deconvolution involves two key objectives: 1) latent image and 2) blur estimation. For latent image estimation, we propose a fast deconvolution algorithm, which uses an image prior of nondimensional Gaussianity measure to enforce sparsity and an undetermined boundary condition methodology to reduce boundary artifacts. For blur estimation, a linear inverse problem with normalization and nonnegative constraints must be solved. However, the normalization constraint is ignored in many blind image deblurring methods, mainly because it makes the problem less tractable. In this paper, we show that the normalization constraint can be very naturally incorporated into the estimation process by using a Dirichlet distribution to approximate the posterior distribution of the blur. Making use of variational Dirichlet approximation, we provide a blur posterior approximation that considers the uncertainty of the estimate and removes noise in the estimated kernel. Furthermore, experiments with synthetic and real data demonstrate that the proposed method is very competitive to the state-of-the-art blind image restoration methods.},
doi = {10.1109/TIP.2015.2478407},
journal = {IEEE Transactions on Image Processing},
number = 12,
volume = 24,
place = {United States},
year = {2015},
month = {9}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record

Citation Metrics:
Cited by: 10 works
Citation information provided by
Web of Science

Save / Share: