A linearized Peaceman-Rachford splitting method with applications to constrained image deblurring problems
Maoying Tian, Min Sun, Shandong Coal · 2016
In this paper, we apply the Peaceman-Rachford splitting method (PRSM) to solve the problem of constrained image deblurring corrupted by Gaussian noise. To speed up PRSM, we linearize its two subproblems and obtain the closed-form solutions. Compared with PRSM, the resulting new method is matrix-inversion free. The global convergence of the new algorithm is proved via the analytic framework of contractive type methods. Numerical comparisons with alternating direction methods (ADMs) illustrate that our proposed algo- rithm is efficient and promising. Keywords. Linearized Peaceman-Rachford splitting method; Image deblurring; Global convergence.