Table 1.

Quantitative comparison of reconstructions by approximant, DNN-L and DNN-S based on a test set of 500 images. Each entry takes the form ’average ± standard deviation’.

	Average PCC ± std.dev		Average PSNR ± std.dev (dB)		Average SSIM ± std.dev
	p = 1	p = 10	p = 1	p = 10	p = 1	p = 10
Approximant fˆ∗	0.148 ± 0.070	0.182 ± 0.086	8.448 ± 4.182	8.465 ± 4.190	0.231 ± 0.111	0.233 ± 0.112
DNN-L output fˆLF	0.812 ± 0.126	0.878 ± 0.083	16.520 ± 2.693	18.439 ± 2.811	0.878 ± 0.088	0.923 ± 0.063
DNN-S output fˆ (q = 0.1)	0.847 ± 0.111	0.891 ± 0.078	17.596 ± 2.612	18.653 ± 2.289	0.903 ± 0.078	0.928 ± 0.065
DNN-S output fˆ (q = 0.2)	0.857 ± 0.088	0.895 ± 0.079	17.816 ± 2.821	18.716 ± 2.286	0.906 ± 0.063	0.928 ± 0.058
DNN-S output fˆ (q = 0.3)	0.859 ± 0.105	0.896 ± 0.074	18.017 ± 2.583	18.749 ± 2.228	0.910 ± 0.075	0.932 ± 0.065
DNN-S output fˆ (q = 0.4)	0.865 ± 0.104	0.895 ± 0.073	18.234 ± 2.484	19.040 ± 2.284	0.926 ± 0.069	0.934 ± 0.057
DNN-S output fˆ (q = 0.5)	0.869 ± 0.112	0.897 ± 0.073	18.600 ± 2.297	19.072 ± 2.271	0.929 ± 0.081	0.935 ± 0.056
DNN-S output fˆ (q = 0.6)	0.869 ± 0.108	0.898 ± 0.077	18.566 ± 2.540	19.041 ± 2.264	0.926 ± 0.080	0.935 ± 0.060
DNN-S output fˆ (q = 0.7)	0.864 ± 0.125	0.895 ± 0.079	17.827 ± 2.377	19.032 ± 2.267	0.927 ± 0.086	0.932 ± 0.069
DNN-S output fˆ (q = 0.8)	0.845 ± 0.115	0.893 ± 0.076	17.577 ± 2.546	19.031 ± 2.755	0.902 ± 0.081	0.931 ± 0.063
DNN-S output fˆ (q = 1)	0.821 ± 0.147	0.890 ± 0.078	17.051 ± 2.306	18.841 ± 2.717	0.902 ± 0.092	0.931 ± 0.063
DNN-S output fˆ (q = 2)	0.819 ± 0.113	0.882 ± 0.078	16.822 ± 2.586	18.645 ± 2.860	0.889 ± 0.081	0.928 ± 0.059