Sparse Image Restoration Using Iterated Linear Expansion of Thresholds

We focus on image restoration that consists in regularizing a quadratic data-fidelity term with the standard l1 sparse-enforcing norm. We propose a novel algorithmic approach to solve this optimization problem. Our idea amounts to approximating the result of the restoration as a linear sum of basic thresholds (e.g. soft-thresholds) weighted by unknown coefficients. The few coefficients of this expansion are obtained by minimizing the equivalent low-dimensional l1-norm regularized objective function, which can be solved efficiently with standard convex optimization techniques, e.g. iterative reweighted least square (IRLS). By iterating this process, we claim that we reach the global minimum of the objective function. Experimentally we discover that very few iterations are required before we reach the convergence.


Published in:
Proceedings of the 18th IEEE International Conference on Image Processing, 1905-1908
Presented at:
18th IEEE International Conference on Image Processing, Brussels, Belgium, September 11-14, 2011
Year:
2011
ISBN:
1522-4880
Keywords:
Laboratories:




 Record created 2013-11-21, last modified 2018-03-17

Postprint:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)