PROCEEDINGS OF THE ITE ANNUAL CONVENTION
Online ISSN : 2424-2292
Print ISSN : 1343-1846
ISSN-L : 1343-1846
2009
Session ID : 17-11
Conference information

17-11 Fundamental study on a denoising method with wavelet color shrinkage : Statistical property of each wavelet layer and its utilization
Nobuhiro FUJIITakashi KOMATSUTakahiro SAITO
Author information
CONFERENCE PROCEEDINGS FREE ACCESS

Details
Abstract
Previously, we have proposed the three different l^p-l^2 color shrinkage schemes, for p=0, 1 and 2, utilizing inter-channel cross-correlations, and have shown that the l^1-l^2 color shrinkage and the l^1-l^2 color shrinkage improve performance of the color-image denoising in the redundant Haar wavelet domain. Superiority or inferiority in performance of the three color shrinkage schemes depends on prior distributions of true wavelet coefficients to be resotored. To improve color-image denoising in the redundant Haar wavelet domain, this paper studies marginal prior distributions of wavelet coefficients of natural clean color images, and addresses the propriety of the three color shrinkage schemes.
Content from these authors
© 2009 The Institute of Image Information and Television Engineers
Previous article Next article
feedback
Top