遥感影像像素级融合方法比较研究

Research on the fixel fusion methed of remote sensing image

遥感影像数据的融合对于利用影像进行的分类、特征提取和目标识别具有重要的意义。文中阐述了IHS彩色空间变换融合法、主成分分析法(PCA)、Brovey法及Gram-schmidt法的算法实现,在此基础上对QuickBird全色波段和多光谱波段进行融合实验,最后从信息熵、灰度均值、相关系数、标准差和视觉效果5个方面综合进行定量与定性评价。分析结果表明,IHS整体上清晰,色调协调,保留了较多的空间信息,细节特征明显,质量较好。

The fusion of remote sensing image is very significant for image classification, feature abstraction and target recognition. In this paper, the algorithms＇ implementation of IHS transform, principal component analysis （PCA）, Brovey transform and Gram-schmidt transform have been studied methodically. And then, panchromatic band and multi-spectral band of QuickBird are used to assess the quality of these fusion algorithms. Finally, for assessing the quality of these images after fusion, information entropy, the gray scale mean value, correlation coefficient, standard deviation and vision effect are generally applied to quantitative assessment and qualitative evaluation. The result indicates that IHS transform has a harmony hue, better quality andpreserve more spatial details.