多光谱与高分辨率图像融合的新算法  

New fusion algorithm for multispectral and high-resolution panchromatric images

作　　者：白蕊[1] 杨万海[1] 张艳妮[1] 

机构地区：[1]西安电子科技大学电子工程学院,陕西西安710071

出　　处：《西安电子科技大学学报》2008年第4期696-702,共7页Journal of Xidian University

基　　金：国家自然科学基金资助(60472085)

摘　　要：提出了一种多光谱与高分辨率图像融合的双树复数小波分维算法.充分利用双树复数小波变换良好的方向选择性,对多光谱的I分量和全色图像进行小波分解,进而在不同子带中进行图像融合,低频采用基于区域能量的融合算法,高频定义不同的分形分维作为权系数进行融合,最后将得到的灰度融合图像进行线性拉伸并替代原来的I分量,得到最终的融合图像.实验结果表明,与经典的IHS变换和小波变换相比,该算法光谱扭曲度仅为前两者的30.31%和50.66%.A Multispectral and High resolution Panchromatic image fusion algorithm based on the DT- CWT transform and fractal dimension is developed. Firstly, the panchromatic images and the component I of the multi spectral image are decomposed to the domain of the DT-CWT transform which has the advantage of good directions, and the image fusion is then implemented in different subbands. The region energy fusion algorithm is adopted as fusion rules in low-pass subbands and high-pass subbands, and images are merged with different fractal dimension as weight. Finally, the stretched grayscale fused image replaces the original intensity component, and the final fused image is achieved. Compared to the traditional HIS transform and wavelet based algorithm, the warp of the proposed algorithm is only 30. 31% and 50. 66% the formers, respectively.

关 键 词：图像融合 双树复数小波变换 分维 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]