Table 1 Quantitative comparison of different dehazing methods on Dense-Haze, NH-Haze, and I-Haze datasets.

From: SAD-Net: a full spectral self-attention detail enhancement network for single image dehazing

Method

Dense-Haze

NH-Haze

I-Haze

SOTS-Indoor

SOTS-Outdoor

PSNR

SSIM

PSNR

SSIM

PSNR

SSIM

PSNR

SSIM

PSNR

SSIM

DehazeNet

13.84

0.43

16.62

0.52

–

–

19.82

0.821

24.75

0.927

AOD-Net

13.14

0.41

15.40

0.57

17.28

0.52

20.51

0.816

24.14

0.919

GridDehazeNet

14.96

0.53

13.80

0.54

18.73

0.77

32.16

0.983

30.86

0.981

MSBDN

15.37

0.49

19.23

0.71

19.62

0.62

32.77

0.981

33.48

0.982

FFA-Net

14.39

0.45

19.87

0.69

19.72

0.73

36.39

0.988

33.57

0.984

DeHamer

16.62

0.56

20.66

0.68

–

–

36.63

0.988

35.18

0.986

AECR-Net

15.80

0.46

19.88

0.71

–

–

37.17

0.990

–

–

MB-TaylorFormer

16.66

0.56

–

–

–

–

42.64

0.994

38.09

0.991

SAD-Net

17.16

0.55

20.48

0.72

21.65

0.84

41.71

0.993

37.78

0.990

  1. We compare PSNR and SSIM, where higher values of PSNR and SSIM indicate better performance. The symbol “–” denotes that the number is not available. Bold indicates the best results.
Back to article page