Piecewise Smooth Segmentation with Sparse Prior
Yutong Li, Yuping Duan · 2018
Exploiting sparsity in the image gradient magnitude has proved effective in preserving sharp edges and reducing noises for many image processing tasks. Based on observation, we build up a novel piecewise smooth segmentation model by utilizing a generalized total variation (TV) prior with p-th power for and a l1data fidelity. We present an efficient algorithm based on the alternating direction method of multipliers (ADMM), where all subproblems can be solved by either one-step Gauss-Seidel iteration or the closed-form solution. Numerical experiments show that the proposed model can achieve more accurate segmentation results than the classical TV based segmentation model.