ISBN: 978-981-11-3671-9 DOI: 10.18178/wcse.2017.06.048
Fast Overlapping Group Sparsity Total Variation Image Denoising Based on Fast Fourier Transform and Split Bregman Iterations
Abstract— The total variation denoising model is considered to be one of the best denoising models.
However, the total variation model always introduces stair-case artifacts. To overcome the drawback, we use
an overlapping group sparsity total variation instead of total variation denoising model. By introducing fast
Fourier transform and split Bregman iteration framework, we propose a fast algorithm to solve the
overlapping group sparse model. Experiments are carried out to compare with the traditional TV denoising
method and state-of-the-art total generalized variation method. The experiments demonstrate that our
algorithm avoids stair-case of traditional TV model.
Index Terms— Total variation, image denoising, fast Fourier transform, overlapping group sparsity.
Yingpin Chen, Zhenming Peng, Xingguo Liu
School of Optoelectronic Information, University of Electronic Science and Technology of China, CHINA
College of Physics and Information Engineering, Minnan Normal University, CHINA
Centers for Biomedical Engineering, University of Science and Technology of China, CHINA
Cite: Yingpin Chen, Linna Wu, Zhenming Peng, Xingguo Liu, "Fast Overlapping Group Sparsity Total Variation Image Denoising Based on Fast Fourier Transform and Split Bregman Iterations," Proceedings of 2017 the 7th International Workshop on Computer Science and Engineering, pp. 278-282, Beijing, 25-27 June, 2017.