Regularized Multiframe Super-Resolution Image Reconstruction Using Linear and Nonlinear Filters

Table 1

The parameter settings of the synthetic dataset in the experiments.

Parameter	Setting value
Number of LR images generated	20 LR images
Lower and upper bound for translations	Randomly between −2 and 2 pixels
Lower and upper bound for rotation angles	Randomly from −0.5  to 0.5
Width of isotropic Gaussian PSF	σ = 0.4
Magnification factor	r = 2
Standard deviation of an additive white Gaussian noise	0.025
Maximum number of iterations	50
Setup image prior for Lorentzian prior:
Regularization weight	λ = 0.0025
Window size	h = 1.8
Scale factor	α = 1
Setup image prior for BTV prior:
Regularization weight	λ = 0.0025
Window size
Scale factor	α = 0.6
Setup image prior for BEP prior:
Regularization weight	λ = 0.0025
Window size	q = 1
Scale factor	α = 1
Threshold parameter	c = 0.2