Collaborative Regularization Models for Color Imaging ProblemsMS68

By considering the gradient of a multichannel image as a 3D tensor with dimensions corresponding to the image domain, spatial derivatives and channels, we introduce collaborative sparsity enforcing norms to address the ill-posedness of color imaging problems. We obtain a regularization framework for color images that uses different channel couplings and enables joint directions of smoothing. We analyze which of the arising models are best suited for color image denoising and other inverse problems.

This presentation is part of Minisymposium “MS68 - Multi-channel image reconstruction approaches
organized by: Jakob Jorgensen (University of Manchester) , Daniil Kazantsev (University of Manchester) .

Catalina Sbert (Universitat de les Illes Balears)
Joan Duran (Universitat de les Illes Balears)
image enhancement, image reconstruction, inverse problems, nonlinear optimization