TY - JOUR T1 - A Primal-Dual Hybrid Gradient Algorithm to Solve the LLT Model for Image Denoising JO - Numerical Mathematics: Theory, Methods and Applications VL - 2 SP - 260 EP - 277 PY - 2012 DA - 2012/05 SN - 5 DO - http://doi.org/10.4208/nmtma.2012.m1047 UR - https://global-sci.org/intro/article_detail/nmtma/5938.html KW - LLT model, image denoising, primal-dual. AB -

We propose an efficient gradient-type algorithm to solve the fourth-order LLT denoising model for both gray-scale and vector-valued images. Based on the primal-dual formulation of the original nondifferentiable model, the new algorithm updates the primal and dual variables alternately using the gradient descent/ascent flows. Numerical examples are provided to demonstrate the superiority of our algorithm.