顾及分割效果的QuickBird影像融合方法对比研究

A Comparative Study of Six Methods for the Fusion of QuickBird Multispectral and Pan Images in Consideration of Segmentation Effect

影像融合是实现高空间分辨率遥感影像中全色数据与多光谱数据优势互补的基本途径,已有的融合方法比较侧重融合图像的视觉效果,较少考虑针对融合影像的分割、分析等后续处理与应用环节。以北京市某地区的QuickBird影像为研究数据,将融合、分割实验研究结果与定量分析相结合,从光谱和几何特征两方面对现有基于像素的主流融合方法IHS、PCA、HPF(高通滤波)、Wavelet-PC(WPC)、Ehlers(ELS)及GS(正交变换)进行对比研究,结果表明,不同融合方法针对同一数据源得到的融合影像在目视效果与定量指标两个方面均存在明显的差异,若顾及后续分割与分析,则以GS融合法的综合效果最佳。

Image fusion is the essential way to integrate the advantages of panchromatic and multi - spectral data of high spatial resolution remote sensing imagery. The comparative study of the present image fusion methods emphasizes the visual effect of the fused imagery and pays less attention to the processing and application steps such as segmentation and analysis which are subsequent to image segmentation. With QuickBird imagery of a place in Beijing as the test data source, the authors combined the quantitative analysis with the study results of fusion and segmentation experiments and performed a comparative study of such pixel - based main fusion methods as IHS, PCA, HPF, Wavelet- PC, Ehlers and GS in the aspects of spectral and geometric features. It is shown that evident differences exist in terms of visual effects and quantitative indices of fused imagery derived from different fusion methods, and that, of these methods, GS seems to be the best.