46
Z. Amini et al.
Table 3.2 Available denoising methods in OCT images [22]
Denoising method
Complex domain
methods
(hardware
methods)
Modification in optical setup
Alternation in incident angle of
the laser beam [5, 84–86]
Alternation in the recording
angle of the back reflected light
[87]
Alternation in the frequency of
the laser beam [88]
Adjustment in imaged subject itself
Weighted averaging schemes
[89]
Registration of multiple frames
by cross correlation [90, 91]
Eye tracking systems [92]
Magnitude domain
methods
Spatial domain
Traditional
methods
Low-pass filtering [93]
2D linear smoothing [1]
Median filter [94–100]
Adaptive wiener filter [71]
Mean filter [71, 101, 102]
Two 1D filters [103]
Advanced methods I-divergence regularization
approach [104]
Non-linear anisotropic filter [27,
34, 35, 62]
Complex diffusion [26, 38]
Directional filtering [36, 37]
Adaptive vector-valued kernel
function [40]
SVM approach [39]
Bayesian estimations [105]
Transform domain Non-Parametric
methods
Sparsity-based denoising [54,
106]
Robust principal component
analysis [41]
Parametric
methods
Wavelet–based methods [42, 43,
51, 62, 63, 107]
Dual tree complex wavelet
transformation [43, 45–47, 69,
108]
Curvelet transform
Circular symmetric Laplacian
mixture model in wavelet
diffusion [44, 65]
None [109–114]