Open AccessTwo anisotropic fourth‐order partial differential equations for image inpainting
Author(s) -
Li Peng,
Li ShuaiJie,
Yao ZhengAn,
Zhang ZuJin
Publication year - 2013
Publication title -
iet image processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.401
H-Index - 45
eISSN - 1751-9667
pISSN - 1751-9659
DOI - 10.1049/iet-ipr.2012.0592
Subject(s) - inpainting , partial differential equation , anisotropic diffusion , discretization , mathematics , diffusion , image (mathematics) , order (exchange) , finite difference method , stability (learning theory) , mathematical analysis , algorithm , computer science , computer vision , physics , finance , economics , machine learning , thermodynamics
In this study, the authors propose two fourth‐order partial differential equations (PDEs) to inpaint the image. By analysing those anisotropic fourth‐order PDEs and comparing their diffusion images, the authors confirm they are forward diffusion or backward diffusion. A numerical algorithm is presented using a finite‐difference method and analyse the stability of discretisation. Finally, they show various experimental results and conclude that the proposed new models are better than the second‐order and third‐order PDEs, especially for weakening the blocky effects.