资源三号卫星湿地多光谱与全色影像融合质量评价

Quality assessment of fusion of ZY-3 multispectral and panchromatic images for coastal wetland areas

从沿海湿地分类效果和定量指标两方面对资源三号(ZY-3)卫星多光谱(MS)图像与全色(PAN)图像融合的质量进行了评价。采用通用明度-色度饱和度(GIHS)变换、à trous小波变换(AWT)、非下采样Contourlet变换(NSCT)和NSCT-GIHS变换结合(用NSCT-GIHS表示)4种融合方法得到了ZY-3 MS-PAN融合图像,并构建了光谱信息指数和空间信息指数定量评价融合效果。定量分析结果表明,采用NSCT-GIHS方法得到的融合图像质量在光谱和空间特征保真度均为最优。基于ZY-3 MS-PAN融合图像,提出了面向对象的湿地分类方法。分类结果表明,NSCT_GIHS融合图像的总体分类精度为最高值96%,Kappa系数为0.9425。可见湿地分类结果与定量评价相一致。

The study was conducted to assess the fusion of the ZY-3 satellite＇ s multi-spectral （MS） images matic （PAN） images from the angles of coastal wetland classification comparison and quantitative MS- and panchro- analysis. Four PAN fusion methods of generalized intensity-hue saturation （GIHS） transformation, a trous wavelet transformation （AWT）, nonsub sampled Contourlet transformation （NSCT） and NSCT-GIHS （NSCT-GIHS combination） fusion quality of the NSCT _ GIHS was superior to that of other three fusion methods from the spectral and spatial fidelity view. Meanwhile, an object-based classification approach was developed for coastal wetland mapping. The classification results demonstrated that NSCT_ GIHS based image obtained the highest overall accuracy and Kappa coefficient, 96% and 0. 9425. It was obvious that wetland classification result was consistent with quantitative assessment.