山区盆地高分一号遥感数据融合方法比较研究  

Comparative Research on Fusion Methods of Remote Sensing Data of GF-1 in Mountain Basin

作　　者：刘祥平 张贞 冯莉莉 黄凌翔 LIU Xiangping;ZHANG Zhen;FENG Lili;HUANG Lingxiang(School of Geology and Geomatics,TCU,Tianjin 300384;School of Mining and Surveying Engineering,HGU,Handan 056038,China)

机构地区：[1]天津城建大学地质与测绘学院,天津300384 [2]河北工程大学矿业与测绘工程学院,河北邯郸056038

出　　处：《天津城建大学学报》2021年第2期137-142,共6页Journal of Tianjin Chengjian University

基　　金：国家自然科学基金(71704128)。

摘　　要：以青海共和盆地作为研究区域,采用Brovey变换、PCA变换、Multiplicative变换和HSV变换4种方法对高分一号的8 m多光谱影像和2 m全色影像进行融合,并对融合结果和分类精度进行评价,筛选出适合研究区地形特征的融合效果的方法.研究结果表明:PCA变换的整体效果最好;Brovey变换次之;HSV变换在空间细节保持度上效果最好但光谱失真严重;Multiplicative变换空间信息损失最大.本研究可为山区盆地区域的高分一号影像融合方法筛选提供一定的实证参考.In this paper,the Gonghe Basin in Qinghai Province was taken as the research area.Four methods,Brovey transform,PCA transform,multiplicative transform and HSV transform were used to fuse the 8m multispectral image and 2m panchromatic image of GF-1.The fusion results and classification accuracy were evaluated,and the better fusion methods suitable for the terrain characteristics of the research area were selected.The results show that PCA transform has the best overall effect,Brovey transform takes the second place,HSV transform has the best effect on spatial detail retention,but the spectrum distortion is serious,and the spatial information loss of multiplicative transform is the largest.This study can provide some empirical reference for the selection of high resolution image fusion methods in mountain basin area.

关 键 词：高分一号卫星 影像融合 质量评价 监督分类 

分 类 号：P237[天文地球—摄影测量与遥感]