基于穗帽变换的IKONOS多光谱和全色影像融合

Fusion of multispectral and panchromatic IKONOS images based on tasseled cap transformation

将IKONOS多光谱影像进行穗帽变换,生成亮度、绿度、湿度等特征波段,然后对IKONOS全色波段与亮度波段进行直方图匹配,用匹配后生成的变量代替穗帽变换生成的亮度波段,再进行逆变换得到融合后的影像.比较了常规融合方法与穗帽融合方法对IKONOS多光谱波段和全色波段影像的融合效果.结果表明,穗帽融合方法较好地将多光谱影像的光谱信息与全色影像的纹理信息融入到融合后影像.目视及定量评价结果都表明,穗帽融合方法优于HIS、Brovey、主成分变换以及加权融合方法.平滑滤波亮度调节(SFIM)方法在光谱信息保真方面表现较好,而穗帽融合方法在纹理信息增加方面表现较好.

A novel IKONOS image fusion method based on tasseled cap （TC） transformation was proposed. The TC method transforms the multispectral IKONOS image to brightness, greenness, and moisture feature variables, and matches the histogram of the IKONOS pan image with that of the brightness image. Then the brightness variable of the TC transformation is replaced by the histogram-matched panchromatic image. Finally, an inverse TC transformation is performed to create a new fused image. Some conventional fusion methods were reviewed and compared with the TC method in terms of fusion quality. The results indicate that the TC method can fuse the spectral information of multispectral image and the texture information of pan image into a new fused image in relative better quality. Both visual and statistical analysis show that the TC method significantly improves the fusion quality compared to the conventional HIS, Brovey, principal component （PC） and weighted fusion techniques. Comparison shows that smoothing filter-based intensity modulation （SFIM） is a superior fusion technique in conserving the spectral information while the TC method is good at improving the spatial detail of fused images.