Open AccessRecursive anisotropic diffusion denoising
Author(s) -
Ostojić V.,
Starčević Đ.,
Petrović V.
Publication year - 2016
Publication title -
electronics letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.375
H-Index - 146
ISSN - 1350-911X
DOI - 10.1049/el.2016.1614
Subject(s) - anisotropic diffusion , noise reduction , enhanced data rates for gsm evolution , anisotropy , diffusion , noise (video) , algorithm , image denoising , reduction (mathematics) , computer science , mathematics , image (mathematics) , artificial intelligence , mathematical optimization , computer vision , optics , geometry , physics , thermodynamics
A recursive anisotropic diffusion algorithm for image denoising is proposed. Adopting the conventional anisotropic diffusion model, extending it to a recursive model leads to better noise removal with the same parameters, and equivalent or improved results with fewer iterations is showed. The reduction in the number of iterations leads to better edge preservation is also demonstrated. Visual and objective performance evaluations both show superior results compared with the classical model. Furthermore, tests on real medical images show successful noise removal with good edge preservation.