Premium
Image restoration through regularization based on error energy minimization
Author(s) -
Pan MinCheng
Publication year - 2010
Publication title -
international journal of imaging systems and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.359
H-Index - 47
eISSN - 1098-1098
pISSN - 0899-9457
DOI - 10.1002/ima.20254
Subject(s) - image restoration , computer science , regularization (linguistics) , algorithm , image processing , image (mathematics) , artificial intelligence , computer vision
Abstract Image restoration has been an important field of image processing for decades. Various methods of restoration have been studied. Meanwhile, regularization presents a very general methodology for image restoration. However, more regularization will remove more noise but will also lose more image detail, and vice versa. In this article, a novel image restoration algorithm is presented, which is derived from regularization by minimizing error energy of noise and ring artifact. Thus, this proposed algorithm is not only immune to noise but also ringing artifact, where a designed high‐pass filter is used in the algorithm proposed here such as the difference of a delta function and a blurring function or an edge operator like a Laplacian operator. This proposed algorithm can achieve high‐resolution images by being implemented with a 1D noiseless signal to reveal its ability of generating high frequencies beyond the diffraction limit and with a 2D computer‐generated noisy image to show the ability of being immune to noise as well as with a 94‐GHz millimeter‐wave image to validate its reliability. © 2010 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 20, 308–315, 2010