Presentation 2001/11/9
Image Restoration under Spatially Correlated Noise
Jun Tsuzurugi, Masato Okada,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, we treat a restoration problem for noisy-degraded image based on the Bayesian inference. As long as we know, it is usually assumed that superimposed noise has no correlation among the pixels. We discuss spatially correlated noise in this paper. The generative statistical models for the original image and the noise are assumed to obey multi-dimensional Gaussian distributions, whose covariance matrixes have translational invariance. We can derive exact description for expectation of the restored image using the Fourier transformation, and found that resultant values of hyper-parameter estimations for minimum error and maximal posterior marginal criteria coincide with each other where generative probabilistic model and that of restoration process belong to the same class, while they do not for the different classes.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Image restoration / Bayesian inference / Hyper-parameter estimation / Fourier transformation
Paper # NC 2001-67
Date of Issue

Conference Information
Committee NC
Conference Date 2001/11/9(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Image Restoration under Spatially Correlated Noise
Sub Title (in English)
Keyword(1) Image restoration
Keyword(2) Bayesian inference
Keyword(3) Hyper-parameter estimation
Keyword(4) Fourier transformation
1st Author's Name Jun Tsuzurugi
1st Author's Affiliation Graduate School of Information Science, Nara Institute of Science and Technology:Institute of Physical and Chemical Research()
2nd Author's Name Masato Okada
2nd Author's Affiliation Institute of Physical and Chemical Research
Date 2001/11/9
Paper # NC 2001-67
Volume (vol) vol.101
Number (no) 432
Page pp.pp.-
#Pages 8
Date of Issue