Learning-Based Dark and Blurred Underwater Image Restoration

Table 2

The test results of objective IQA methods between our algorithm and underwater enhancement and restoration algorithms.

Image ID	Assessment methods	Original image	CLAHE	GC	ICM	Rayleigh Distribution	UCM	RoWS	UDCP	Ours	Ground Truth
100600	PSNR	15.39	16.826	18.110	16.059	9.813	19.071	14.924	14.925	27.036	Inf
100600	SSIM	0.121	0.448	0.574	0.250	0.183	0.426	0.012	0.011	0.840	1
100600	IL-NIQE	85.737	92.681	85.937	82.378	53.607	76.323	157.962	154.894	39.235	28.716
100600	NIQE	14.473	8.559	8.652	8.687	8.810	8.862	20.830	20.901	5.756	5.1267
100602	PSNR	15.839	18.815	19.633	16.740	10.495	17.191	15.169	14.727	24.579	Inf
100602	SSIM	0.280	0.624	0.646	0.412	0.246	0.358	0.117	0.030	0.842	1
100602	IL-NIQE	83.133	84.796	84.337	72.221	62.313	72.443	85.413	87.961	32.961	34.990
100602	NIQE	7.467	8.458	8.490	8.360	10.198	8.131	7.731	8.021	5.9664	6.849
100603	PSNR	11.717	13.872	14.803	12.363	12.292	12.952	11.353	11.290	23.613	Inf
100603	SSIM	0.205	0.505	0.605	0.349	0.375	0.277	0.089	0.077	0.867	1
100603	IL-NIQE	80.893	80.292	68.379	65.774	45.447	64.708	80.856	84.260	37.503	33.236
100603	NIQE	7.310	7.176	7.296	7.504	9.025	7.085	7.339	7.186	5.622	4.641
100605	PSNR	11.633	13.156	14.822	12.173	11.956	14.140	10.985	11.023	22.167	Inf
100605	SSIM	0.134	0.352	0.512	0.237	0.279	0.357	0.007	0.017	0.770	1.00
100605	IL-NIQE	113.811	96.737	106.721	100.244	62.865	85.374	112.699	105.880	51.919	28.02
100605	NIQE	9.979	9.278	9.248	9.533	9.931	9.300	12.789	10.492	5.4618	4.944