Table 3. Performance comparison with the CARN-M[9] model. The red color shows the best performance and the blue color shows the second-best performance in terms of PSNR and SSIM [23] comparison. 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	MSSSIM / UQI
Set5	×2	37.61 /0.9587	0.9952 /0.9990	37.62 /0.9586	0.9951 /0.9990	37.64 /0.9586	0.9952 /0.9990
	×3	33.94 /0.9228	0.9903 /0.9979	33.92 /0.9227	0.9903 /0.9979	33.91 /0.9226	0.9903 /0.9979
	×4	31.69 /0.8870	0.9815 /0.9966	31.70 /0.8873	0.9817 /0.9966	31.63 /0.8867	0.9815 /0.9965
Set14	×2	33.24 /0.9142	0.9855 /0.9976	33.26 /0.9140	0.9855 /0.9976	33.23 /0.9142	0.9855 /0.9976
	×3	30.04 /0.8356	0.9729 /0.9952	30.05 /0.8355	0.9729 /0.9952	30.08 /0.8364	0.9731 /0.9953
	×4	28.30 /0.7736	0.9531 /0.9930	28.31 /0.7738	0.9531 /0.9930	28.30 /0.7739	0.9532 /0.9929
BSDS100	×2	31.93 /0.8960	0.9818 /0.9970	31.94 /0.8961	0.9818 /0.9970	31.93 /0.8960	0.9818 /0.9970
	×3	28.87 /0.7988	0.9657 /0.9942	28.88 /0.7988	0.9657 /0.9942	28.88 /0.7987	0.9657 /0.9942
	×4	27.35 /0.7278	0.9416 /0.9920	27.36 /0.7280	0.9416 /0.9920	27.37 /0.7281	0.9417 /0.9919
Urban100	×2	31.09 /0.9170	0.9856 /0.9962	31.09 /0.9171	0.9856 /0.9962	31.15 /0.9176	0.9857 /0.9962
	×3	27.35 /0.8331	0.9719 /0.9918	27.34 /0.8331	0.9719 /0.9918	27.39 /0.8339	0.9721 /0.9918
	×4	25.38 /0.7606	0.9488 /0.9876	25.38 /0.7607	0.9487 /0.9876	25.41 /0.7613	0.9491 /0.9876