Multi-scale Total Variation with Automated Regularization Parameter Selection for Color Image Restoration

Abstract

In this talk, we are concerned with the multi-scale total variation model for image restoration. Since the parameter controls the trade-off between the image smoothness and the preservation of details, we consider a spatially dependent choice and propose an iterative method to determine a set of (local) parameters corresponding to the image regions pertinent to different scales. Based on a primal-dual technique, we introduce an efficient algorithm to solve this model in order to restore blurred noisy images. Numerical results show that this method can provide better performance of suppressing noise as well as preserving details in the image.

DOI: 10.1007/978-3-642-02256-2_23

Cite this paper

@inproceedings{Dong2009MultiscaleTV, title={Multi-scale Total Variation with Automated Regularization Parameter Selection for Color Image Restoration}, author={Yiqiu Dong and Michael Hinterm{\"{u}ller}, booktitle={SSVM}, year={2009} }