Super-resolution via a fast deconvolution with kernel estimation

EURASIP Journal on Image and Video Processing

Table 3 Quantitative comparison in terms of RMSE, PSNR and SSIM

Example(factor)	Index	Bicubic	08’TOG [22]	11’IPOL [7]	11’SIAM [8]	14’TIP [9]	Ours
House(2)	RMSE	6.4742	6.0985	5.4645	5.3947	7.4008	5.1384
	PSNR	31.9071	32.4263	33.3798	33.4915	30.7452	33.9143
	SSIM	0.8831	0.8858	0.8961	0.8969	0.8761	0.8958
House(3)	RMSE	8.9995	7.2818	7.8816	7.8106	9.1071	7.1034
	PSNR	29.0465	30.886	30.1985	30.2771	28.9432	31.1015
	SSIM	0.847	0.8655	0.861	0.8615	0.8554	0.872
House(4)	RMSE	11.018	9.1549	9.8511	9.7658	11.7308	9.0703
	PSNR	27.2887	28.8977	28.2611	28.3366	26.7443	28.9784
	SSIM	0.8169	0.8418	0.8302	0.8311	0.8278	0.8444
Race(2)	RMSE	11.18	11.2381	10.3093	10.1244	12.5453	10.8124
	PSNR	27.1619	27.1169	27.8662	28.0234	26.1612	27.4524
	SSIM	0.6912	0.6911	0.7353	0.7386	0.6536	0.692
Race(3)	RMSE	13.9661	13.0151	13.0624	12.9765	14.1891	13.1136
	PSNR	25.2293	25.8418	25.8103	25.8676	25.0917	25.7764
	SSIM	0.6042	0.6078	0.6407	0.6429	0.5955	0.6171
Race(4)	RMSE	15.7714	14.6136	14.8819	14.8098	15.7557	14.6909
	PSNR	24.1734	24.8357	24.6776	24.7198	24.182	24.7898
	SSIM	0.5524	0.562	0.5804	0.5826	0.5575	0.5761