Table 2. Performance comparison with the MDSR [6] model. The red color indicates the best performance while the blue color shows the second-best performance in terms of PSNR and SSIM [26]. We also measure performance with MS-SSIM [27] and UQI [27]

Dataset	scale	Multi-path		Linear single-path		Cubic single-path
		PNSR / SSIM	MSSSIM /UQI	PNSR / SSIM	MSSSIM /UQI	PNSR / SSIM	MSSIM / UQI
Set5	×2	37.86 /0.9594	0.9953 /0.9990	37.87 /0.9595	0.9953 /0.9990	37.85 /0.9595	0.9953 /0.9990
	×3	34.31 /0.9253	0.9908 /0.9980	34.31 /0.9256	0.9908 /0.9980	34.29 /0.9256	0.9908 /0.9980
	×4	32.03 /0.8921	0.9826 /0.9968	32.05 /0.8929	0.9828 /0.9968	32.03 /0.8924	0.9827 /0.9968
Set14	×2	33.49 /0.9165	0.9861 /0.9977	33.54 /0.9167	0.9861 /0.9977	33.53 /0.9170	0.9863 /0.9977
	×3	30.24 /0.8396	0.9738 /0.9954	30.30 /0.8408	0.9740 /0.9955	30.29 /0.8408	0.9739 /0.9954
	×4	28.49 /0.7788	0.9546 /0.9931	28.51 /0.7796	0.9548 /0.9932	28.54 /0.7802	0.9550 /0.9932
BSDS100	×2	32.10 /0.8981	0.9823 /0.9971	32.14 /0.8985	0.9824 /0.9971	32.13 /0.8985	0.9824 /0.9971
	×3	29.04 /0.8027	0.9666 /0.9943	29.07 /0.8033	0.9668 /0.9944	29.07 /0.8035	0.9668 /0.9944
	×4	27.53 /0.7334	0.9433 /0.9922	27.56 /0.7343	0.9436 /0.9922	27.55 /0.7344	0.9436 /0.9922
Urban100	×2	31.77 /0.9246	0.9873 /0.9966	31.90 /0.9260	0.9877 /0.9967	31.89 /0.9261	0.9877 /0.9967
	×3	27.91 /0.8467	0.9749 /0.9925	28.00 /0.8488	0.9753 /0.9926	27.99 /0.8491	0.9754 /0.9926
	×4	25.86 /0.7778	0.9542 /0.9887	25.94 /0.7809	0.9550 /0.9887	25.93 /0.7810	0.9550 /0.9887