摘要
水体信息的高精度提取是水资源监测、调查与管理等研究领域的关键问题。本文选取国产GF-1卫星影像作为数据源,根据影像中典型地物光谱特征采用归一化水体指数法(NDWI)、多波段谱间关系法对研究区进行了水体提取,并通过结果分析提出了基于主成分分析与多尺度分割技术的综合水体信息提取方法。对以上4种方法进行对比分析,验证结果表明:归一化水体指数法易受阴影信息影响;改进的多波段谱间关系法能较为完整地提取水体信息,但受小范围阴影信息影响;主成分分析综合法总体精度较高;多尺度分割提取法能有效地分离水体信息与非水体信息,水体信息提取的效果最佳。
The high accuracy extraction of water information is a key problem in the research of water resources monitoring, investiga- tion and management. We choose the domestic GF - 1 satellite imagery as the data source and the method of normalized water body in- dex （NDWI） and multi - band spectrum - photometric was used to extract the water body of the study area by analyzing the typical fea- tures spectral characteristics of the data. Based on the analysis of the above results, taking the principal component analysis and muhi - scale segmentation method as the indirection the integrated water body information extraction method is proposed. Through the com- parative analysis of the above 4 methods, the verification results indicate that the normalized difference water index method is suscepti- ble to shadow information, the modified multi -band spectrum -photometric method can be used to extract relatively complete water information but affected by a small range of shadow information, the principal component analysis method has a higher overall preci- sion, and the multi - scale segmentation and extraction method can be used to effectively separate water information from non - water information and achieve the best results of extracting water information.
出处
《测绘与空间地理信息》
2017年第10期70-73,76,共5页
Geomatics & Spatial Information Technology
基金
大屯矿区生态与水文地质环境遥感信息研究(20123702002534)
海洋测绘重点实验室开放基金(2013B01)资助
关键词
GF-1
水体信息提取
NDWI
主成分分析
多尺度分割
GF -1
water information extraction
NDWI
principal component analysis
multi -scale segmentation
