Skip to main content

Table 2 Performance comparison of Neiye, Japan

From: Land cover classification combining Sentinel-1 and Landsat 8 imagery driven by Markov random field with amendment reliability factors

Method

Precision

Urban

Sand

Vegetation

Farmland

Overall accuracy

Kappa

MRF-ARF

PA

90.67%

74.01%

69.16%

90.56%

87.86%

0.7219

UA

96.45%

65.65%

65.12%

68.20%

MRF-RF

PA

54.33%

74.26%

85.20%

81.38%

60.69%

0.3905

UA

99.67%

22.32%

24.91%

86.23%

ATWT-EMD

PA

42.38%

62.10%

91.54%

90.59%

52.05%

0.3168

UA

98.13%

15.11%

26.65%

78.74%

SAR only

PA

80.25%

59.66%

53.28%

65.41%

75.31%

0.5149

UA

98.92%

26.16%

29.75%

75.92%

Optical only

PA

58.73%

64.97%

87.74%

96.07%

64.89%

0.4250

UA

96.91%

19.68%

59.68%

53.72%