Digital Path Approach Despeckle Filter for Ultrasound Imaging and Video

Table 2

Comparison of the filtering algorithms applied for standard test videos corrupted with different noise scenarios.

Video sequence noise σ^2	Foreman			Salesman			Tennis			Fetus
	0.2	0.4	0.6	0.2	0.4	0.6	0.2	0.4	0.6
PSNR results (dB)
Noisy	30.23	18.70	12.52	33.83	21.83	15.42	31.13	19.20	12.65	23.17
Wiener 2D	33.87	24.00	19.70	33.00	25.94	19.70	29.93	23.92	16.61	24.12
Wiener 3D	33.90	24.24	20.00	32.75	26.18	20.00	29.56	24.12	17.57	26.86
TGauss	30.60	23.30	20.61	33.56	26.50	20.61	27.56	22.50	17.32	26.30
Median 3D	30.00	26.65	24.51	29.15	27.48	24.51	21.70	21.21	19.64	26.36
NLM2D	35.05	27.92	22.28	34.37	26.88	22.28	25.96	26.39	17.75	29.10
NLM3D	35.67	28.02	22.34	32.79	27.06	22.34	30.75	27.64	17.81	30.28
BM3D	34.44	30.93	23.12	35.25	32.30	24.92	31.97	28.63	22.64	28.40
FDPA	32.74	25.32	20.23	31.28	27.24	20.24	27.40	24.09	16.73	24.80
STFFDPA	33.47	23.09	18.25	33.37	26.16	18.24	28.91	23.06	15.06	24.51
EPF2D	33.86	25.69	20.99	32.16	27.64	24.07	28.50	24.36	19.89	25.76
EPF1st3D	34.39	28.05	25.02	34.40	29.28	25.02	30.85	27.20	20.99	28.92
EPFlast3D	33.27	28.10	25.05	34.28	29.07	25.06	30.11	25.85	21.20	29.34
MSSIM results
Noisy	0.72	0.25	0.09	0.90	0.48	0.22	0.82	0.35	0.16	0.40
Wiener 2D	0.89	0.47	0.41	0.90	0.69	0.41	0.78	0.55	0.24	0.49
Wiener 3D	0.89	0.47	0.43	0.90	0.70	0.43	0.78	0.56	0.26	0.64
TGauss	0.81	0.40	0.41	0.92	0.67	0.41	0.82	0.49	0.25	0.61
Median 3D	0.84	0.63	0.59	0.86	0.76	0.59	0.59	0.47	0.35	0.75
NLM2D	0.92	0.71	0.50	0.91	0.73	0.50	0.85	0.68	0.30	0.90
NLM3D	0.92	0.71	0.50	0.88	0.73	0.50	0.86	0.69	0.31	0.91
BM3D	0.90	0.83	0.52	0.93	0.89	0.60	0.81	0.69	0.48	0.86
FDPA	0.87	0.55	0.42	0.88	0.74	0.42	0.72	0.54	0.24	0.54
STFFDPA	0.89	0.44	0.33	0.92	0.69	0.33	0.79	0.49	0.21	0.52
EPF2D	0.90	0.58	0.34	0.90	0.77	0.57	0.78	0.59	0.35	0.64
EPF1st3D	0.91	0.72	0.64	0.93	0.83	0.63	0.86	0.69	0.40	0.88
EPFlast3D	0.89	0.71	0.60	0.93	0.82	0.60	0.84	0.65	0.41	0.89