Image Denoising Using Nonlocal Means with Shape-Adaptive Patches and New Weights
Table 1
The PSNR results by different denoising methods. In each cell, the results of the four denoising methods are presented in the following order: top left, NLM [21]; top right, NLM-SAP [29]; bottom left, BM3D-SAPCA [41]; bottom right, our proposed denoising method.
σ
20
35
50
100
C. man
29.85
29.75
27.09
26.98
25.37
24.82
21.42
20.86
30.91
29.97
28.17
27.54
26.59
26.02
22.87
22.50
House
32.40
32.62
29.79
29.00
27.74
26.37
23.16
22.79
33.89
33.28
31.37
30.90
29.52
28.97
25.08
24.77
Peppers
30.17
30.64
27.08
27.50
25.16
25.26
21.02
21.02
31.57
30.82
28.75
28.22
26.98
26.51
23.24
22.90
Lena
31.58
31.97
28.95
29.00
27.39
27.12
23.93
23.99
33.20
32.62
30.72
30.28
29.06
28.67
25.36
25.37
Barbara
30.47
30.50
27.60
26.83
25.72
24.69
22.14
22.02
32.37
31.36
29.61
28.34
27.68
26.40
23.22
22.65
Boats
29.80
29.64
26.99
26.80
25.28
25.04
22.19
22.31
31.02
30.34
28.51
27.96
26.88
26.47
23.68
23.42
Man
29.75
29.58
27.03
26.92
25.41
25.39
22.54
22.95
30.83
30.32
28.38
28.04
26.93
26.59
23.96
23.98
Hill
29.77
29.45
27.14
26.92
25.49
25.55
22.84
23.28
30.85
30.32
28.61
28.21
27.19
26.86
24.26
24.29
Average
30.47
30.51
27.70
27.49
25.94
25.53
22.40
22.40
31.83
31.12
29.26
28.68
27.60
27.06
23.95
23.73
The bold values represent the best results among the four methods, which has been explained in the second paragraph of Section 4.
