Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/17522
Title: Bayesian Approaches For Image Restoration
Authors: YU WEIMIAO
Keywords: Image Restoration, Edge Preservation, Kernel Learning, Bayesian Apporach, Rotational Blurring
Issue Date: 27-Sep-2007
Source: YU WEIMIAO (2007-09-27). Bayesian Approaches For Image Restoration. ScholarBank@NUS Repository.
Abstract: The problem of recovering discontinuous signal (1D/2D) from its degraded and noisy version will be studied in this dissertation. Discontinuities encode important, crucial and significant information, however, the removal of the noise and the preservation of the existed discontinuities are conflicting. Bayesian model select is robust, however expensive due to the a??curse of dimensionalitya??. We provide an approach to reduce the calculation burden in Bayesian model select. Some simplifications of the calculation for the normalization constants make it efficient. We suggest that the so-called posterior evidence should be used for the model select. Bayesian Kernel based on the sparse kernel learning is presented for the both 1D and 2D signals. The given cost function only has one global minimum; therefore the problem of ill-conditioness is successfully solved. The Bayesian approach for the blur identification in spatial domain is proposed. This approach can identify extend of the Point Spread Function and its parameter. A discrete formulation of compact coordinates transform is proposed. The concept of compactness is the first time to be addressed and the optimization on the compactness is studied. The experiments and simulations show our method can solve the spatial variant problem of rotational blurred image restoration.
URI: http://scholarbank.nus.edu.sg/handle/10635/17522
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
BAYESIAN APPROACHES FOR IMAGE RESTORATION.pdf21 MBAdobe PDF

OPEN

NoneView/Download

Page view(s)

255
checked on Dec 11, 2017

Download(s)

88
checked on Dec 11, 2017

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.