Accelerated GNHSS iterative method for weighted Toeplitz regularized least-squares problems from image restoration

作者：Cui, Jing-Jing^*; Peng, Guo-Hua; Lu, Quan; Huang, Zheng-Ge

来源：COMPUTATIONAL & APPLIED MATHEMATICS, 2018, 37(5): 6152-6175.

DOI：10.1007/s40314-018-0681-8

• HSS iteration method
• Preconditioning
• Image restoration
• Least squares problems
• Weighted Toeplitz matrices

For fast solving weighted Toeplitz least-squares problems from image restoration, we establish an accelerated GNHSS (AGNHSS) method based on the Hermitian and skew-Hermitian splitting. The convergence of the new iteration method is established theoretically and its quasi-optimal iteration parameters are discussed. It is seen that the AGNHSS method converges faster when proper parameters are chosen. In particular, the convergence conditions of the AGNHSS method, the optimal parameters and some comparisons between AGNHSS and NSHSS are given when it is used for solving the linear image restoration problems. Meanwhile, the spectral properties of the preconditioned matrix are investigated. Experimental results further demonstrate that the new method is effective and confirm that our theoretical results are correct.

• 出版日期2018-11
• 单位西北工业大学