1: \begin{abstract}

2: We propose a new fast algorithm for solving one of the standard

3: approaches to ill-posed linear inverse problems (IPLIP),

4: where a (possibly non-smooth) regularizer is minimized

5: under the constraint that the solution explains the observations

6: sufficiently well. Although the regularizer and constraint are

7: usually convex, several particular features of these problems

8: (huge dimensionality, non-smoothness) preclude the use of

9: off-the-shelf optimization tools and have stimulated a considerable

10: amount of research. In this paper, we propose a new efficient

11: algorithm to handle one class of constrained problems (often

12: known as basis pursuit denoising) tailored to image recovery

13: applications. The proposed algorithm, which belongs to the family

14: of augmented Lagrangian methods, can be used to deal with a variety

15: of imaging IPLIP, including deconvolution and reconstruction from

16: compressive observations (such as MRI), using either total-variation

17: or wavelet-based (or, more generally, frame-based) regularization.

18: The proposed algorithm is an instance of the so-called

19: {\it alternating direction method of multipliers}, for which

20: convergence sufficient conditions are known; we show that these

21: conditions are satisfied by the proposed algorithm. Experiments on

22: a set of image restoration and reconstruction benchmark problems

23: show that the proposed algorithm is a strong contender for the

24: state-of-the-art.

25: \end{abstract}

26: