Non‐convex nonlocal adaptive tight frame image deblurring

The challenge of the image restoration is to recover more detailed information from degraded images. Based on observations that wavelet frames have efficient representation ability details and nonconvex regularization in model may admit unbiased solutions, this paper, order details, a ℓ p ( 0 < 1 $\ell _p(0<p<1$ ) established combing with nonlocal adaptive mean doubly augmented Lagrangian (MDAL) method context model. Specifically, MDAL proposed solve Furthermore, mitigate trade-off between error noise magnification error, spatially thresholding methods based filter are introduced algorithm. convergence analysis algorithm also obtained using KL inequality. Numerical experiments demonstrate has competent deblurring denoising ability, comparable state-of-the-art methods.