一种改进的HSI图像融合算法  

An Improved HSI Image Fusion Algorithm

作　　者：刘建伟[1] 侯翔文[1] 

机构地区：[1]西安工业大学理学院,西安710021

出　　处：《西安工业大学学报》2013年第5期368-371,共4页Journal of Xi’an Technological University

摘　　要：针对传统HSI(Hue-Saturation-Intensity,HSI)图像融合算法中出现颜色失真现象,提出了一种改进的基于HSI变换的多光谱图像和高空间分辨率图像融合的新算法.该算法首先对多光谱图像进行HSI变换,其次对HSI变换后的强度分量进行高通滤波处理,接着与全色图像进行过加权融合处理,得到新的强度分量,其中加权系数从客观评价标准逆推得到,最后进行HSI逆变换,得到新的融合图像.通过与传统的HSI算法对比评价,多组实验结果表明本文提出的融合算法得到的融合图像的信息熵和平均梯度(清晰度)都增加了,而且多光谱图像的光谱偏差指数变小了.因此该算法在提高多光谱图像的空间细节表现能力和保持光谱信息上都有较好的效果,同时还具有较好的通用性.The paper describes a new algorithm for the fusion of low-resolution multi-spectral image and high-resolution panchromatic image to solve the problem of color distortion caused by the traditional HSI（Hue-Saturation-Intensity）transformation method.First,the multi-spectral image is transformed with HSI.Then the intensity component of the multi-spectral image is processed by the high-pass filter,and it is merged with panchromatic image by over weighted average（OWA）,with the weighted coefficient is obtained from the image fusion evaluation criterion.Finally,the fusion image is obtained by inverse HSI transformation.The results show that,compared with the traditional HSI method the information entropy and the average gradient of the fusion image obtained by the proposed method increase,while the optical spectral bias index decreases.It is concluded that the proposed algorithm can enchance spatial resolution of fusion image,obtain more details and feature information,and preserve spectral information of the original multi-spectral image well.It has great versatility.

关 键 词：图像融合 HSI 过加权 高通滤波 熵 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]