Volume 434, Number 2, May I 2005
|Page(s)||795 - 800|
|Published online||11 April 2005|
Multiple-image deblurring with spatially-variant point spread functions
Chip Computers Consulting s.r.l., Viale Don L. Sturzo 82, S.Liberale di Marcon, 30020 Venice, Italy e-mail: firstname.lastname@example.org
2 ESA-VILSPA, Apartado 50727, 28080 Madrid, Spain
3 Department of Mathematics and Computer Science, Emory University, Atlanta, GA 30322, USA e-mail: email@example.com
4 ESA-VILSPA, Apartado 50727, 28080 Madrid, Spain e-mail: firstname.lastname@example.org
Accepted: 3 January 2005
We generalize a reliable and efficient algorithm, to deal with the case of spatially-variant PSFs. The algorithm was developed in the context of a least-square (LS) approach, to estimate the image corresponding to a given object when a set of observed images are available with different and spatially-invariant PSFs. Noise is assumed additive and Gaussian. The proposed algorithm allows the use of the classical LS single-image deblurring techniques for the simultaneous deblurring of the observed images, with obvious advantages both for computational cost and ease of implementation. Its performance and limitations, also in the case of Poissonian noise, are analyzed through numerical simulations. In an appendix we also present a novel, computationally efficient deblurring algorithm that is based on a Singular Value Decomposition (SVD) approximation of the variant PSF, and which is usable with any standard space-invariant direct deblurring algorithm. The corresponding Matlab code is made available.
Key words: methods: data analysis / methods: statistical / techniques: image processing
© ESO, 2005
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.