一种改进的Curvelet变换遥感影像融合方法

提出了一种改进的Curvelet变换的融合算法,以IKONOS影像为样本数据,融合过程中将原全色波段与多光谱影像中未参与HIS分解的波段进行加权组合,形成新全色分量,再将此分量与明度分量分别进行第二代Curvelet变换,以形成新明度分量,HIS逆变换后形成融合影像。并与改进的HIS变换融合、传统的HIS变换与Curvelet变换结合的融合算法相比较,结果表明,改进的方法能增加融合影像的信息量,降低光谱扭曲度,提高融合影像与原多光谱影像的相似度。

A Method for Remote Sensing Image Fusion Based on Improved Curvelet Transform

This paper presents a method for image fusion that combines the improved HIS algorithm with theCurvelet transform algorithm to fuse the IKONOS imagery.Firstly,add the multispectral band that is nottransformed by the HIS algorithm into the panchromatic band with powers to obtain a new panchromaticcomponent.Secondly,use the second generation Curvelet to decompose the new component and the intensitycomponent,then fuse the Curvelet coefficients based on the local energy fusion rule in the corresponding scales,and finally obtain the fused image through reconstructing the fused coefficients.For proving the superiority of thismethod,several parameters are used to evaluate the result comparing with other results which are obtainedthrough the methods of HIS algorithm,improved HIS algorithm as well as HIS and Curvelet integratedalgorithm.The evaluation results show that the proposed method can increase the information entropy,decreasethe spectrum distortion degree of the fused image,and improve the structural similarity between the fused imageand the original multispectral image,and the whole results show that the integrated method has a practical value.