A hierarchically sampling global sparse transformer in data stream mining for lightweight image restoration

EURASIP Journal on Advances in Signal Processing

Table 4 Performance comparison of image deblurring

Method	GoPro [51]		HIDE [52]		RealBlur-R [53]		RealBlur-J [53]
Method	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
Xu et al. [54]	21.00	0.741	–		34.46	0.936	27.14	0.830
DeblurGAN [55]	28.70	0.858	24.51	0.871	33.79	0.903	27.97	0.834
Nah et al. [51]	29.08	0.914	25.73	0.874	32.51	0.841	27.87	0.827
Zhang et al. [56]	29.19	0.931	–		35.48	0.947	27.80	0.847
DeblurGAN-V2 [25]	29.55	0.934	28.61	0.875	35.26	0.944	28.70	0.866
SRN [57]	30.26	0.934	28.36	0.915	35.66	0.947	28.56	0.867
Shen et al. [52]	–		28.89	0.930	–		–
Gao et al. [58]	30.90	0.935	29.11	0.913	–		–
HST	30.94	0.934	29.29	0.913	33.93	0.943	27.63	0.867