Table 1. Free parameters of the proposed method and their values, which have been used in all experiments.

Value	Parameter name	Parameter group
$Ma{x}_{Iter}=40$	Number of iterations	EM convergence criteria (Algorithm 4)
$\alpha =0.56$	The ratio of entropy of each window w.r.t. the entropy of the whole image	Calculation of $\mathrm{\Delta }\tau$
$W_{min}=\mathrm{m}\mathrm{a}\mathrm{x}(7,{\displaystyle \frac{1}{20}min\left(width,height\right))}$	Minimum size of sliding window
$W_{max}=\mathrm{m}\mathrm{a}\mathrm{x}(19,{\displaystyle \frac{1}{8}min\left(width,height\right))}$	Maximum size of sliding window
$\beta =1.3$	Coefficient of selecting outlier transformation matrix
$\rho =2$	Penalty amount	Finding the optimal value of Δτ by the ADMM method
${\epsilon }^{tollerance}={10}^{-7}$	Threshold of determination of convergence condition
$\epsilon =0.15$	Merge threshold	Number of MOG models in estimating distortion and noise
$Ma{x}_k=10$	Number of primary Gaussian models
$m=25$	Number of data choices	RANSAC method
$C=10$	Number of iterations
${\epsilon }^{Ransac}=$ 0.5	Similarity threshold for selecting support set
λ = 0.0001	Regularization parameter	WALS method