1: \begin{abstract}

2: The \emph{alternating direction method of multipliers} (ADMM) has recently

3: sparked interest as a flexible and efficient optimization tool for imaging inverse

4: problems, namely deconvolution and reconstruction under non-smooth convex

5: regularization. ADMM achieves state-of-the-art speed by

6: adopting a \emph{divide and conquer}

7: strategy, wherein a hard problem is split into simpler, efficiently solvable

8: sub-problems ({\it e.g.}, using fast Fourier or wavelet transforms, or

9: simple proximity operators). In deconvolution, one of these

10: sub-problems involves a matrix inversion ({\it i.e.}, solving a linear system),

11: which can be done efficiently (in the discrete Fourier domain) if

12: the observation operator is circulant, {\it i.e.}, under periodic boundary

13: conditions. This paper extends ADMM-based image deconvolution to the more

14: realistic scenario of unknown boundary, where the observation operator

15: is modeled as the composition of a convolution (with arbitrary boundary

16: conditions) with a spatial mask that keeps only pixels that do not depend

17: on the unknown boundary. The proposed approach also handles,

18: at no extra cost, problems that combine the recovery of missing

19: pixels ({\it i.e.}, inpainting) with deconvolution.

20: We show that the resulting algorithms inherit

21: the convergence guarantees of ADMM and illustrate its performance on

22: non-periodic deblurring (with and without inpainting of interior pixels)

23: under total-variation and frame-based regularization.

24: 

25: \end{abstract}