非下采样剪切波变换域下遥感图像融合  

Remote sensing image fusion in the non-subsampled shearlet transform domain

作　　者：熊璋玺 刘明亮[1,3] 郭擎 李安[2] XIONG Zhangxi;LIU Mingliang;GUO Qing;LI An(School of Electronic Engineering, Heilongjiang University, Harbin 150080, China;Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, China;Key Laboratory of Information Fusion Estimation and Detection, Harbin 150080, China)

机构地区：[1]黑龙江大学电子工程学院,黑龙江哈尔滨150080 [2]中国科学院空天信息研究院,北京100101 [3]黑龙江省信息融合估计与检测重点实验室,黑龙江哈尔滨150080

出　　处：《哈尔滨工程大学学报》2022年第2期290-297,共8页Journal of Harbin Engineering University

基　　金：国家自然科学基金项目(61771470);黑龙江省自然科学基金项目(LH2020F046).

摘　　要：多光谱图像与全色图像的融合已被广泛应用于提高后续图像处理效果以满足图像进一步应用的需要,本文提出了一种在非下采样剪切波变换域中基于区域清晰度加权和导向滤波相结合的遥感图像融合方法。利用剪切波变换将多光谱图像的亮度分量与全色图像分别分解为低频子带和高频子带;针对高低频子带的特点分别设计高低频子带融合规则;对融合系数取剪切波逆变换和亮度-色度-饱和度逆变换得到融合图像。与6种典型融合方法通过9种包含光谱和空间质量评价的指标进行比较,实验结果表明:本文方法在主观视觉效果和客观评价方面都达到了极好效果,在提升空间分辨率的同时有效的保留了光谱信息。Image fusion of multispectral and panchromatic images has been widely applied to improve the effectiveness of subsequent image processing to address the need for further image application.A new image fusion method based on regional sharpness weighting and guided filtering in the non-subsampled shearlet transform(NSST)domain is proposed in this paper.First,the luminance component of the multispectral and panchromatic images is respectively decomposed into low-and high-frequency sub-bands by the NSST.Then,the fusion rules of high-and low-frequency sub-bands are respectively designed in accordance with the characteristics of high-and low-frequency sub-bands.Finally,the fusion image is obtained by taking the inverse NSST and IHS transform.Compared with six typical fusion methods through nine indicators,including spectral and spatial quality evaluation,the experimental results show that the proposed method achieves optimal performance considering the subjective visual effect and objective assessment and effectively preserves the spectral information while improving spatial resolution.

关 键 词：图像融合 多光谱图像 全色图像 区域清晰度加权 导向滤波 非下采样剪切波变换 

分 类 号：P2[天文地球—测绘科学与技术]