Speckle Noise Removal Model Based on Diffusion Equation and Convolutional Neural Network

Table 3

PSNR results of different models on test images.

Test image	I	II	III	IV	V	VI	VII	VIII	Average
Noise level σ = 2
Proposed	29.74	31.89	31.00	31.67	29.31	31.30	33.64	30.59	31.14
SARD	26.15	28.43	27.93	28.10	26.43	28.53	30.01	27.71	28.23
VA	26.32	28.96	28.01	28.31	26.77	28.69	30.28	27.82	28.51
DnCNN	29.65	31.88	30.95	31.66	29.23	31.24	33.63	30.57	31.10
RED-net	28.51	30.39	29.14	29.97	28.07	29.73	32.07	29.14	29.63
Noise level σ = 3
Proposed	27.42	30.03	29.10	29.66	27.30	29.37	32.05	28.79	29.21
SARD	26.78	27.54	26.07	25.78	23.12	25.65	27.51	24.17	25.61
VA	24.11	25.74	25.48	25.98	23.77	25.87	27.98	24.76	25.91
DnCNN	27.36	29.85	29.09	29.64	27.14	29.30	31.94	28.77	29.14
RED-net	26.75	28.87	27.27	28.28	26.43	28.20	30.39	27.78	28.06
Noise level σ = 4
Proposed	25.82	28.43	27.72	28.39	25.97	27.88	30.58	27.52	27.78
SARD	23.13	25.41	24.83	25.51	23.17	24.82	27.32	24.76	25.01
VA	23.51	25.62	25.14	25.86	23.67	25.21	27.88	25.03	25.62
DnCNN	25.65	28.33	27.78	28.12	25.81	27.62	30.40	27.48	27.64
RED-net	30.31	27.94	27.08	27.61	25.48	27.05	29.53	27.00	27.14