An linearized alternating direction method for total variation image restoration

Total variation (TV) regularization is popular in image reconstruction due to its edge-preserving property. In this paper, we propose to solve the TV regularized problem by linearized alternating minimization method based on the augmented Lagrangian function at each iteration and a Barzilai-Borwein (BB) stepsize rule. Numerical results show that the linearized alternating direction method can restore the degraded image with high quality, good convergence and stability.