Algorithm 1: Pseudo-code of the proposed zoom quality metric
	Input: Zoom photo I, over-completer DCT dictionary D, mean and variances of pristine MVG parameters ${\mu }_y$, ${\Sigma }_y$, weighting parameters $k_1$, w and l.
1	Initialization:
2	Compute the photo gradient $\nabla I$ using Sobel operator;
3	Partition the photo I into non-overlapping 96 x 96 patches ${\mathbf{x}}_k$
4	The measurement of sharpness:
5	foreach $k=1,2,\dots ,N$do
6		Solve the sparse coding coefficients ${\mathbf{\alpha }}_k^{*}$ using the OMP algorithm [73];
7		Calculate the patch variance ${\sigma }_k^2$;
8		Sort and select the top $l\%$ patches according to the variance ${\sigma }_k^2$;
9	end$\phantom{(}$
10	Compute mean KL-divergence or energy: ${\sum }_{k=1}^{N_s}\frac{{({\mathbf{\alpha }}_k^{*})}^T{\mathbf{\alpha }}_k^{*}}{{\sigma }_k^2}$
11	Compute the residual gradient image: $\nabla I-\nabla I^{\prime }$
12	Compute entropy of the residual: $\mathbb{E}(log\frac{1}{P(S\odot (|\nabla I-\nabla I^{\prime }|)|\mathbf{\alpha })})$
13	The sharpness index: $SS={\sum }_{k=1}^{N_s}\frac{{({\mathbf{\alpha }}_k^{*})}^T{\mathbf{\alpha }}_k^{*}}{{\sigma }_k^2}+k_1·\mathbb{E}(log\frac{1}{P(S\odot (|\nabla I-\nabla I^{\prime }|)|\mathbf{\alpha })})$
14	The measurement of naturalness:
15	foreach $k=1,2,\dots ,N$ do$\phantom{(}$$\phantom{(}$
16		Compute the MSCN map of I using (20)
17		Compute the GGD and AGGD parameters of each patch using (21), (23), (24)
18	end$\phantom{(}$
19	Estimate the ${\mu }_x$ and ${\Sigma }_x$ through (25)
20	The naturalness index:  $NS=\sqrt{{({\mu }_x-{\mu }_y)}^T{\left(\frac{{\Sigma }_x+{\Sigma }_y}{2}\right)}^{-1}({\mu }_x-{\mu }_y)}$  Output: Zoom quality metric $Q=SS+w·NS$