A robust anisotropic diffusion filter with low arithmetic complexity for images

EURASIP Journal on Image and Video Processing

Table 2 Performance for low-density salt and pepper noise

Methods	Test image	NI	Noise type − S&P Noise % − 20		Noise type − S&P Noise % − 70		Noise type-GN + S&P Noise % − 0.1 + 20		Noise type-GN + S&P Noise % − 0.1 + 70
Methods	Test image	NI	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
RF λ = 0.25	Cameraman	5	19.413	0.8673	12.5457	0.7796	16.906	0.8159	12.4372	0.7782
	Lena	5	20.4622	0.9693	12.6513	0.8751	12.9513	0.8622	13.0337	0.8639
	Pepper	5	19.7905	0.9622	13.2004	0.8719	17.2988	0.9096	12.9513	0.8622
AMD λ = 0.25	Cameraman	5	26.079	0.9958	20.5091	0.8927	19.178	0.8270	19.7392	0.8033
	Lena	5	28.7007	0.9964	18.0798	0.9520	16.4546	0.8962	18.2120	0.9057
	Pepper	5	28.6385	0.9923	18.6443	0.9462	19.5751	0.9199	16.4546	0.8962
AADF λ = 0.25	Cameraman	1	40.1719	0.9997	28.7566	0.9967	14.4454	0.8380	10.5868	0.7873
	Lena	1	41.9936	0.9999	26.0650	0.9975	15.9324	0.9441	9.3239	0.8923
	Pepper	1	42.2684	0.9999	31.4408	0.9999	14.4793	0.9380	10.8085	0.9021
FORADF λ = 0.25	Cameraman	5	34.8502	0.9997	24.4484	0.9947	19.2064	0.8296	17.6850	0.8265
	Lena	5	37.7526	0.9998	27.0534	0.9962	18.1284	0.9212	19.9176	0.9315
	Pepper	5	37.5130	0.9998	25.5219	0.9942	19.5065	0.9260	18.1284	0.9212
FORADF λ = 1	Cameraman	5	28.1563	0.9925	23.916	0.9771	20.080	0.8377	18.822	0.8328
	Lena	5	31.2451	0.9990	26.2455	0.9951	19.3510	0.9271	20.5072	0.9381
	Pepper	5	31.7647	0.9986	25.1724	0.9883	20.7433	0.9342	19.3510	0.9271