基于导向滤波的多光谱智能融合方法研究  

Research on Multi-Spectral Intelligent Fusion Method Based on Guided Filtering

作　　者：陈应霞[1] 陈艳[1] CHEN Ying-xia;CHEN Yan(School of Computer Science,Yangtze University,Jingzhou Hubei 434023,China)

机构地区：[1]长江大学计算机科学学院,湖北荆州434023

出　　处：《计算机仿真》2025年第2期280-285,310,共7页Computer Simulation

基　　金：2022年中国高校产学研创新基金项目(2021RYC06002);2022年度湖北省自然资源厅自然资源科技项目(MB1644959/2021-70118);2022年湖北省教育厅科研计划项目(B2022040)。

摘　　要：锐化(Pan-sharpening)的目的是通过合理的规则融合全色图像(panchromatic,PAN)和低空间分辨率的多光谱图像(low multispectral image,LMS),从而获得高空间高光谱分辨率的多光谱图像(high multispectral image,HMS)。为此提出了一种新的基于导向滤波和组合差分进化优化(combined differential evolution,CoDE)的锐化方法。方法首先利用导向滤波技术对原始多光谱图像(multispectral image,MS)进行边缘增强处理;接着,采用智能优化方法之组合差分进化算法对提出的融合模型进行优化。从主观视觉和客观指标两个维度上,采用两组滑坡灾害影像数据验证了新方法的有效性和高效性。提出的新方法,能为滑坡灾害识别、救援和灾后重建提供数据服务。The purpose of Pan-sharpening is to fuse panchromatic images(PAN)and low spatial resolution multispectral images(LMS)by reasonable rules,so as to obtain high spatial and high spectral resolution multispectral images(HMS).This paper proposed a noval Pan-sharpening method based on guided filtering and combined differential evolution optimization(CoDE).Firstly,the edge of the original multispectral image(MS)is enhanced by guided filtering technology.Then,the combined differential evolution algorithm is used to optimize the proposed fusion model.Finally,through subjective vision and objective evaluation,the effectiveness and efficiency of the proposed method are verified by two sets of landslide geological hazard image,thus providing data services for landslide disaster identification,rescue and post-disaster reconstruction.

关 键 词：锐化 全色图像 多光谱图像 导向滤波 组合差分进化优化 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]