谐波分析法高光谱影像融合及其光谱信息保真度评价

摘要

Combined with the Hyperion hyperspectral image and ALI high spatial resolution band of the EO-1 satellite,the paper puts forward the harmonic analysis fusion (HAF) algorithm of hyperspectral image and the derivative spectral d-value's information entropy (DSD-IE) model of the spectral-fused information fidelity evaluation.Through calculating and evaluating some parameters such as the DSD-IE values,average gradient and standard deviation of the sample spectra meanwhile compared with the fused hyperspectral images by the traditional methods like the principal component analysis (PCA),Gram-Schmidt and wavelet,the fused hyperspectral iamge by the HAF has proved to have the higher information degree of spatial integration and spectral fidelity,and the better superiorities in the reliability,accuracy and applicability.%提高高分辨率影像之间的空间信息融入度和光谱信息保真度是当今高光谱影像融合研究的热点问题.本文结合EO-1卫星的Hyperion高光谱影像与ALI高空间分辨率波段,针对最大限度地克制光谱信息失真与光谱形态变化等问题,提出了高光谱影像的谐波分析融合(HAF)算法和融合影像光谱信息保真度评价的导数光谱差值信息熵(DSD-IE)模型.通过计算和评价采样像元样本的导数光谱差值曲线信息熵、平均梯度和标准差等参量,同时比较主成分分析(PCA)、Gram-Schmidt (GS)和小波分析等方法的高光谱融合影像,证实了基于HAF的高光谱影像融合结果具有较高的空间信息融入度和光谱信息保真度,在空间信息和光谱信息的可靠性、准确性和实用性等方面都明显优于其他传统的高光谱影像融合方法.