An Improvement of BM3D Image Denoising and Deblurring Algorithm by Generalized Total Variationстатья

Информация о цитировании статьи получена из Scopus, Web of Science
Дата последнего поиска статьи во внешних источниках: 11 апреля 2019 г.

Работа с статьей

[1] Nasonov A., Krylov A. An improvement of bm3d image denoising and deblurring algorithm by generalized total variation // 2018 7th European Workshop on Visual Information Processing (EUVIP). — 2018. — P. 1–4. In this work we propose a post-processing method for BM3D algorithm that has become a state-of-the-art image denoising and deblurring algorithm. Although BM3D algorithm produces results with high objective metrics values, it also adds noticeable high-frequency artifacts. We suppress these artifacts using second order Total Generalized Variation (TG V) algorithm. TGV algorithm is an extension of Total Variation denoising method but it does not tend to make images piecewise constant. We also suggest an efficient numerical scheme for TGV minimization. In order to validate the proposed idea, tests were performed on noisy real images and synthetic images with different levels of noise. [ DOI ]

Публикация в формате сохранить в файл сохранить в файл сохранить в файл сохранить в файл сохранить в файл сохранить в файл скрыть