IKONOS影像融合方法对比研究及应用效果评价  

Comparative study on the fusion methods and application effect of the IKONOS image

作　　者：李莹莹[1] 洪金益[1] 张建国[2] 尹展[2] 黄梅[1] 田浩[3] 

机构地区：[1]中南大学,长沙410000 [2]有色金属矿产地质调查中心,北京100012 [3]西华师范大学,南充637002

出　　处：《矿产勘查》2014年第3期499-504,共6页Mineral Exploration

基　　金：甘肃省矿山开发遥感调查与监测项目(编号:121201120041)资助

摘　　要：文章以IKONOS影像为数据源,使用常用的4种融合方法:IHS、主成分变换、Brovey和Pansharp,对影像的全色及多光谱波段进行融合,并对各方法的融合影像进行了目视对比、统计分析和光谱分析,发现与原始影像相比,融合影像的纹理信息、光谱信息在Pansharp融合影像中保持得最好。选用相同的训练样本,对原始影像和融合影像应用支持矢量机的方法(SVM),分为裸地、煤堆、河流、矿山建筑物、道路等5类地物,并对分类结果进行了精度评价。在4种融合影像中,Pansharp融合影像的分类精度最高,总体分类精度达到82.89%。综合分析得出,Pansharp方法是一种较好的融合方法。Based on IKONOS images as data source, this article introduces the image fusion for the multispectral bands and panchromatic band by means of four usual methods, i.e. , IHS, principal component analysis, Brovey and Pansharp, then the visual contrast, statistical and spectral analysis for each fused image were followed. By comparing with the original images, it is found that the texture and spectral information is best preserved in the Panshrap image. Using the same training samples and the technique of SVM, the feature classification was carried out in the original and the fused images. Total of five kinds features were identified as followings ： bare ground, coal heaps, rivers, mine buildings and roads, and the classification accuracy were all assessed. Of the four kinds of fused images, the accuracy of the fused Pansharp image is the best,in which the accuracy reaches 82.89%. Comprehensively, the fusion method of Pansharp is better than the other three methods.

关 键 词：IKONOS影像 融合方法 光谱信息 纹理信息 影像分类 

分 类 号：P237[天文地球—摄影测量与遥感]