Image Super-Resolution Network Based on Feature Fusion Attention

Table 2

Benchmark tests results, average PSNR/SSIM, bold is the best result, and italic is the second best result.

Method	Set14			BSD100			Urban100
	X2	X3	X4	X2	X3	X4	X2	X3	X4
Bicubic	30.240	27.550	26.000	29.560	27.210	25.960	26.880	24.460	23.140
	0.869	0.774	0.703	0.843	0.739	0.668	0.840	0.735	0.658
SRCNN	32.450	29.300	27.500	31.360	28.410	26.900	29.500	26.240	24.520
	0.907	0.822	0.751	0.888	0.786	0.710	0.895	0.799	0.722
LapSRN	33.080	29.790	28.190	31.800	28.820	27.320	30.410	27.070	25.210
	0.913	0.832	0.772	0.895	0.797	0.728	0.910	0.827	0.755
SRDenseNet	—	—	28.500	—	—	27.530	—	—	26.050
	—	—	0.778	—	—	0.734	—	—	0.782
CARN	33.520	30.290	28.600	32.090	29.060	27.580	31.920	28.060	26.070
	0.917	0.841	0.781	0.898	0.803	0.735	0.926	0.849	0.784
EDSR	33.920	30.520	28.800	32.320	29.250	27.710	32.930	28.800	26.640
	0.920	0.846	0.788	0.901	0.809	0.742	0.935	0.865	0.803
RDN	34.010	30.570	28.810	32.340	29.260	27.720	32.890	28.800	26.610
	0.921	0.847	0.787	0.902	0.809	0.742	0.935	0.865	0.803
SwinSR	33.070	32.182	31.091	33.345	28.900	31.474	33.856	28.793	24.525
	0.891	0.889	0.847	0.933	0.825	0.854	0.959	0.874	0.781
DeFiAN	33.789	33.747	31.087	32.955	29.050	31.448	34.196	28.537	25.050
	0.908	0.913	0.847	0.931	0.830	0.870	0.959	0.866	0.797
RDAN	35.815	33.881	31.228	33.281	30.133	29.646	34.289	28.816	26.473
	0.923	0.908	0.850	0.939	0.859	0.838	0.958	0.885	0.827