Performance of Alternating Direction Method of Multipliers(ADMM) Based on Deconvolving Images with Unknown Boundaries
|T.Nithya1, Dr.S.Leela Lakshmi2
|Related article at Pubmed, Scholar Google|
Deconvolution is an ill-posed inverse problem, it can be solved by imposing some form of regularization(prior knowledge) on the unknown blur and original image. This formulation allows both frame-based or total-variation regularization. In several imaging inverse problems, ADMM is an efficient optimization tool that achieves state-of-the-art speed, by splitting the underlying problem into simpler, efficiently solvable sub-problems. In deonvolution the observation operator is circulant under periodic boundary conditions, one of these sub-problems requires a matrix inversion, which can be efficiently computable(via the FFT). we show that the resulting algorithms inherit the convergence guarantees of ADMM. These methods are experimentally illustrated using frame-based regularization; the results show the advantage of our approach over the use of the ―edge taper‖ function (in terms of improvement in SNR).