【摘要】 遥感影像像素级融合是遥感信息分析与处理过程中十分常用的处理方法之一,但目前该领域缺乏相应的理论研究。通过对GIF和GCOS模型的对比分析,建立遥感数据像素级融合的统一理论框架,并给出几种具有代表性的融合方法在统一框架下的表达。在此基础上,例证统一理论框架在算法等价性判别、融合性能理论分析、融合算法设计等方面的作用。在此框架下的研究表明,Brovey变换融合法与基于三角形模型的IHS方法完全等价,PCA方法为GS方法的特例。此外还揭示了组合现有方法以提高融合性能的原理。这些研究结果说明统一理论框架在问题分解方面的合理性和进行遥感影像像素级融合理论分析的有效性。更多还原

【Abstract】 Image fusion on pixel level is a common process for information analysis of remotely sensed data. However, there is a lack of theoretical research in the field. Based on comparative analysis of GIF and GCOS model, the paper proposed a unified theoretical framework of remotely sensed image fusion. Description of several typical fusion methods are provided. Moreover, works of algorithm equivalence discriminant, fusion performance analysis and fusion algorithm development in the framework are provided. Some interesting and valuable conclusions are drawn by analysis in the unified theoretical framework. The result shows that BT is equivalent to IHS based on triangle model, and PCA is a special case of GS. The reason why hybrid methods could improve fusion performance is also revealed. The study proves the validation and effectivity of the framework on theoretical analysis of image fusion.更多还原