摘要
为更加客观公正地比较融合结果、评估融合算法及优化融合过程,该文在分析主要遥感图像融合质量评价方法的基础上,将定量评价指标分为三类:评价图像亮度信息的指标、评价空间信息保持能力的指标和评价光谱信息保持能力的指标。最后,通过仿真试验,以光学遥感图像融合为例,验证了4种常用的像素级融合算法的适用性和有效性。IHS变换法扭曲了源图像的光谱特性,容易产生光谱退化;Wavelet变换法在光谱特性保持方面具有优势,但容易出现分块效应和模糊现象;PCA变换法较多地保留了源图像的细节纹理和结构特征,但会失去源图像的部分物理特性;缨帽变换法融合结果地物边缘清晰,但对细微光谱信息的保持能力较弱。研究表明,这3类指标可以作为遥感图像融合的客观效果评价准则,并为融合结果的后续应用提供借鉴意义。
In order to compare fusion effects,evaluate fusion algorithms and optimize fusion process,three quantitative evaluation indexes were presented on the basis of analyzing quality evaluation methods of remote sensing image fusion：imagery brightness index,spatial information index,and spectral information index.Finally,optical remote sensing imagery fusion was taken for example.The applicability and effectiveness of four pixels level fusion algorithms were verified by means of simulation experiment.IHS transformation distorted the spectral characteristics of imagery,so the feature spectra was easily to become vestigial;Wavelet transformation had superiority on maintaining spectral characteristics,but the result easily became block and blurred;PCA transformation maintained most detail texture and architectural feature of source imagery,but it couldn＇t maintain detail spectral information.The research shows that,these three indexes can be taken as the evaluation criterion of remote sensing image fusion,they can also be used as good reference for many applications.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2011年第10期95-100,共6页
Transactions of the Chinese Society of Agricultural Engineering
基金
"高分辨率对地观测系统专项"项目
国家自然科学基金项目(41101312)
中国博士后科学基金项目(20110490197)
关键词
遥感
图像融合
模型
质量评价
空间信息
光谱信息
remote sensing
image fusion
models
quality evaluation
spatial information
spectral information
