An Efficient Universal Noise Removal Algorithm Combining Spatial Gradient and Impulse Statistic

Table 6

Comparative restoration results in PSNR (dB) and MAE (the second row) for mixed noise (𝑝=20%sp, 𝜎=10: salt-and-pepper and Gaussian; 𝑝=20%rv, 𝜎=10: random-valued impulse and Gaussian; 𝑝=30%: salt-and-pepper and random-valued impulse noise).

Method	Baboon image			Barbara image
	𝑝=20%sp, 𝜎=10	𝑝=20%rv, 𝜎=10	𝑝=30%	𝑝=20%sp, 𝜎=10	𝑝=20%rv, 𝜎=10	𝑝=30%
Noisy image	11.98	15.25	13.98	12.30	15.92	14.17
	21.04	15.32	16.39	14.35	9.000	13.75
3 × 3 median filter	20.43	21.01	21.76	23.06	23.62	23.17
	8.754	8.396	5.215	5.197	5.433	3.881
5 × 5 median filter	20.59	20.66	20.90	22.81	22.86	22.79
	8.369	8.419	6.368	5.275	5.674	4.422
Bilateral filter	19.81	21.00	19.80	21.70	22.65	21.43
	9.612	8.404	5.958	8.006	7.822	6.120
SDROM	21.23	22.19	24.07	25.29	25.58	25.71
	7.860	7.403	4.256	4.143	4.420	3.329
ROAD-TRIF	21.71	22.32	24.13	25.66	25.94	25.66
	7.403	7.314	4.221	3.800	4.205	3.410
ROLD-EPR	21.38	22.26	24.03	25.48	26.02	25.88
	7.614	7.329	4.261	3.961	4.229	3.230
RORD-WMF	21.87	22.32	24.30	25.79	26.11	25.94
	7.363	7.300	4.062	3.787	4.268	2.978
DWM	21.34	22.15	23.89	25.44	25.65	25.30
	7.681	7.411	4.402	4.005	4.404	3.664
SBF	21.79	22.20	24.06	25.72	25.87	25.75
	7.404	7.387	4.257	3.802	4.339	3.323
SG-DARD-TRIF	22.41	22.37	24.15	26.71	26.15	26.03
	6.983	7.293	4.079	3.625	4.123	2.619