珞珈一号融合多源数据的建成区提取  

Extraction of Built-up Area Based on Luojia 1-01 Combined with Multi Source Data

作　　者：唐霞 汤军[1] 李外宾 高贤君[1] 宋树华[1,2,3] TANG Xia;TANG Jun;LI Waibin;GAO Xianjun;SONG Shuhua(School of Earth Sciences,Yangtze University,Wuhan 430100,China;School of Mathematics and Statistics,Lingnan Normal University,Zhanjiang,Guangdong 524048,China;KQ GEO Technologies Co.Ltd.,Beijing 102600,China)

机构地区：[1]长江大学地球科学学院,武汉430100 [2]岭南师范学院数学与统计学院,广东湛江524048 [3]苍穹数码技术股份有限公司,北京102600

出　　处：《遥感信息》2023年第1期78-87,共10页Remote Sensing Information

基　　金：自然资源部地理国情监测重点实验室开放基金项目(2020NGCM07)。

摘　　要：针对使用夜间灯光影像提取建成区存在灯光溢出、过饱和等所导致提取精度受限的问题,采用空间分辨率更高的珞珈一号影像,分别结合高精度的兴趣点、道路、归一化差异植被指数和地温数据,构建多个夜间灯光修正指数,引入最大类间方差算法优化动态阈值法来提取建成区,以武汉市为例对建成区提取精度进行研究。结果表明,综合多源数据的城市建成区提取效果普遍要优于单一数据源,但并非融合的数据类型越多建成区提取精度就越高。使用几何平均值法融合珞珈一号、兴趣点和地温数据构建的LPALUI指数提取建成区最精确,可保留丰富的边界和内部空间信息,其总体精度、Kappa系数和F1-score高达93.7%、0.608、0.957,是城市建成区提取的理想多源数据指数。In order to solve the problem that the extraction accuracy of built-up areas was limited due to light overflow and oversaturation when using night light images to extract built-up areas,this paper used Luojia 1-01 image with higher spatial resolution,combined with high-precision points of interest(POI),roads,normalized differential vegetation index(NDVI)and land surface temperature(LST)data to construct the multiple night light correction indexes.The OTSU algorithm was introduced to optimize the dynamic threshold to extract built-up area.Taking Wuhan as an example,the extraction accuracy of built-up area was studied.The results show that the effect of urban built-up area extraction based on multi-source data is generally better than that based on single data source.However,it is wrong that the more data types are combined,the higher extraction accuracy of the built-up area.LPALUI index constructed by Luojia 1-01,POI and LST data using the geometric average method is the most accurate for extracting the built-up area,which can retain rich boundary and internal spatial information,and has great applicability.Its overall accuracy,Kappa coefficient and F1-score are 93.7%、0.608 and 0.957,which is an ideal multi-source data index for extracting urban built-up area.

关 键 词：珞珈一号 建成区提取 多源数据 阈值分割 最大类间方差算法OTSU 

分 类 号：P237[天文地球—摄影测量与遥感]