Skip to main content

Table 1 The denoising results in dB for six test images with noise power in the range [5,100] gray values

From: An augmented Lagrangian multi-scale dictionary learning algorithm

σ/PSNR

'Barbara'

'House'

'Boat'

'Lena'

'Peppers'

'Cameraman'

5/34.15

37.91

39.50

37.89

38.13

38.08

38.22

 

37.74

39.32

37.81

37.97

37.78

37.93

10/28.13

34.19

36.05

33.58

34.30

34.59

34.00

 

33.94

35.97

33.39

34.02

34.17

33.71

15/24.61

32.10

34.46

31.33

32.10

32.54

31.67

 

31.92

34.31

31.17

31.86

32.27

31.35

20/22.11

30.63

33.35

29.78

30.58

31.03

30.27

 

30.55

33.18

29.65

30.38

30.76

29.97

25/20.17

29.33

32.18

28.60

29.45

29.88

29.04

 

29.25

32.10

28.52

29.30

29.70

28.78

50/14.15

24.57

27.76

24.98

25.83

26.20

25.73

 

24.65

27.85

24.99

25.76

26.19

25.65

75/10.61

21.50

25.19

22.67

23.55

23.69

23.38

 

21.52

25.26

22.72

23.56

23.72

23.38

100/8.13

20.19

23.46

21.66

22.02

21.90

21.70

 

20.26

23.40

21.67

22.01

21.86

21.72

  1. For each test setting, two results are provided: our ALM-DL algorithm (top); and K-SVD algorithm (bottom). T he best result in each set is highlighted.