基于高分可见光遥感指数的城市阴影高效提取研究
唐晔,刘小燕,崔耀平,史志方,邓亮,陈准

Research on Efficient Extraction of Urban Shadow based on High-resolution Visible Light Remote Sensing Index
Ye TANG,Yaoping CUI,Xiaoyan LIU,Zhifang SHI,Zhun CHEN,Liang DENG
表1 各精度分析区在NSVDI、USI、深度学习和OUSI的精度评价
Table 1 Shadow detection accuracy evaluation of the NSVDI index, USI index, deep learning, and OUSI index in the six accuracy analysis areas
方法/研究区生产者精度用户精度总体精度
阴影非阴影阴影非阴影
ηs/%ηs/%ρs/%ρs/%τ/%
NSVDI89.6550.0063.4083.3369.49
92.5047.5063.7986.3670.00
66.6750.0057.1060.0058.30
83.3376.6678.1282.1480.00
63.3373.3370.3766.6668.30
90.0080.0081.8188.8885.00
USI86.2093.3392.5987.5089.83
80.0097.5096.9682.9788.75
63.3386.6682.6170.2775.00
86.6696.6696.2987.8791.66
66.6696.6695.2374.3581.66
76.66100.00100.0081.0888.33
深度学习82.7590.0088.8884.3786.40
72.5087.8790.0076.5981.25
46.6696.6693.3364.4471.66
83.3393.3392.5984.8488.33
70.0096.6695.4576.3183.33
66.6690.0086.9572.9778.33
OUSI89.6593.3392.8590.3291.52
82.50100.00100.0085.0091.25
73.3396.6695.6578.3785.00
96.6696.6696.6696.6696.66
76.6696.6695.8380.5586.66
83.33100.00100.0085.7191.66