Abstract

We describe a simple approach useful for improving noisy, blurred images. Our approach is based on the use of a parallel block-based low rank factorization technique for projection based reduction of matrix dimensions and on a customized iteratively reweighted CG approach followed by the use of a Fourier Wiener filter. The regularization scheme with a transform basis offers variable residual penalty and increased per-iteration performance. The outlined approach is particularly aimed at high blur and noise settings.

This work is licensed under a Creative Commons Attribution 4.0 License.