Dou Wen, Sun Hong-quan, Chen Yun-hao
Transportation College, Southeast University, Nanjing 210096, China.
Guang Pu Xue Yu Guang Pu Fen Xi. 2011 Mar;31(3):746-52.
Remotely sensed image fusion is a critical issue, and many methods have been developed to inject features from a high spatial resolution panchromatic sensor into low spatial resolution multi-spectral images, trying to preserve spectral signatures while improving spatial resolution of multi-spectral images. However, no explicit physical information of the detection system has been taken into account in usual methods, which might lead to undesirable effects such as severe spectral distortion. Benefiting from the proper decomposition of the image fusion problem by a concise image fusion mathematical model, the present paper focuses on comparing reasonable modulation coefficient of spatial details based on analysis of the spectral response function (SRF). According to the classification of former methods, three modulation coefficients based on SRF of sensors were concluded, which lead to three image fusion methods incorporating spatial detail retrieved by Gaussian high-pass filter. All these methods were validated on Ikonos data compared to GS and HPM method.
遥感图像融合是一个关键问题，人们已经开发出许多方法，将高空间分辨率全色传感器的特征注入到低空间分辨率多光谱图像中，试图在提高多光谱图像空间分辨率的同时保留光谱特征。然而，常规方法没有考虑检测系统的明确物理信息，这可能会导致诸如严重光谱失真等不良影响。得益于简洁的图像融合数学模型对图像融合问题的合理分解，本文重点基于光谱响应函数（SRF）分析比较空间细节的合理调制系数。根据以往方法的分类，得出了基于传感器SRF的三种调制系数，进而产生了三种结合高斯高通滤波器检索到的空间细节的图像融合方法。与GS和HPM方法相比，所有这些方法都在Ikonos数据上进行了验证。
