Efficient Minimization Methods of Mixed l2-l1 and l1-l1 Norms for Image Restoration
Authors:
- Haoying FuHFHaoying Fu
- Michael K. NgMNMichael K. Ng
- Mila NikolovaMNMila Nikolova
Open Access
- 1 January 2006
- journal article
- Published by Society for Industrial & Applied Mathematics (SIAM) in SIAM Journal on Scientific Computing
- Vol. 27 (6) , 1881-1902
Abstract
Image restoration problems are often solved by finding the minimizer of a suitable objective function. Usually this function consists of a data-fitting term and a regularization term. For the least squares solution, both the data-fitting and the regularization terms are in the $\ell$2 norm. In this paper, we consider the least absolute deviation (LAD) solution and the least mixed norm (LMN) solution. For the LAD solution, both the data-fitting and the regularization terms are in the $\ell$1 norm. For the LMN solution, the regularization term is in the $\ell$1 norm but the data-fitting term is in the $\ell$2 norm. Since images often have nonnegative intensity values, the proposed algorithms provide the option of taking into account the nonnegativity constraint. The LMN and LAD solutions are formulated as the solution to a linear or quadratic programming problem which is solved by interior point methods. At each iteration of the interior point method, a structured linear system must be solved. The precondit...Keywords
This publication has 41 references indexed in Scilit:
- Image Sharpening by Flows Based on Triple Well PotentialsJournal of Mathematical Imaging and Vision, 2004
- A Preconditioner for Generalized Saddle Point ProblemsSIAM Journal on Matrix Analysis and Applications, 2004
- A property of the minimum vectors of a regularizing functional defined by means of the absolute normIEEE Transactions on Signal Processing, 1997
- Regularization of linear least squares problems by total bounded variationESAIM: Control, Optimisation and Calculus of Variations, 1997
- A Sparse Approximate Inverse Preconditioner for the Conjugate Gradient MethodSIAM Journal on Scientific Computing, 1996
- An algorithm for the minimization of mixed l/sub 1/ and l/sub 2/ norms with application to Bayesian estimationIEEE Transactions on Signal Processing, 1994
- Digital filters as absolute norm regularizersIEEE Transactions on Signal Processing, 1992
- Restoration of images with missing high-frequency components by minimizing the L_1 norm of the solution vectorApplied Optics, 1985
- Restoration of images with missing high-frequency components using quadratic programmingApplied Optics, 1983
- Digital image restoration using quadratic programmingApplied Optics, 1980