Well-Posed Inhomogeneous Nonlinear Diffusion Scheme for Digital Image Denoising
Citations
89 citations
71 citations
63 citations
31 citations
Cites background or methods from "Well-Posed Inhomogeneous Nonlinear ..."
...(f) Prasath and Singh [22] — Edge detector based Anisotropic Diffusion (EAD) ∂u ∂t = div α(x)∇u 1 + |∇u|2 /K 2 ...
[...]
...Adaptive schemes comparison results on Kikis128 × 128 synthetic image, (in each sub-figure, the right image shows the surface form of the left image): (a) EED [46]; (b) CED [47]; (c) SAD [48]; (d) ATV [16]; (e) ALD [19]; (f) EAD [22]; (g) original image and its surface form given for comparison; (h) proposed scheme with nonlinear diffusion (WWBF)....
[...]
...There exists various choices [17,19,21,22] for theweight functionα in Eq....
[...]
...5,1](Gσ ⋆∇u(x, t))/(1+|Gσ ⋆ ∇u|2 /K 2) No diffusion at edges Noise remain along edges [19] 1 + Mcχc ,Mc ≫ 0 constant Edge indication Excessive blurring [22] 1 − Gσ ⋆ Canny(u(x, t)) Retains multi-scale edges Cannot handle high noise [37] exp −Θ(Var(2)Nx (u(x, t)), θ)/δ Contextual discontinuities Stippled pattern artifacts [38] 1I + 1Ic exp − (|Gσ ⋆ ∇u(x)| /K)(2) Handles impulse noise Cannot handle textures...
[...]
...Note that the PSNR values are closer together when adaptive fidelity is used (SAD, ATV, ALD, EAD, and our WWBF) in Table 2, but MSSIM values indicate a better performance of the proposed approach....
[...]
31 citations
References
28,073 citations
15,225 citations
12,560 citations
5,769 citations
3,447 citations