摘要
为解决Landsat 8 OLI多光谱图像和CBERS-04 P5全色图像的融合问题,设计了一种基于WT-MB变换的融合方法;以宁夏回族自治区吴忠市的图像融合为案例,从定量评价、目视判读评价和SVM分类精度评价3个方面比较WT-MB变换方法与GS、PCA、MLT、Brovey变换方法的融合效果,然后以总体精度和Kappa系数评判融合图像的分类精度.实验结果表明:WT-MB变换方法在增强融合OLI图像的空间结构信息的同时,更有效地保持原OLI图像的光谱信息,提高了图像分类精度,能够更准确地服务于环境遥感监测.
The WT-MB transform image fusion method was designed to solve the integration of the Landsat 8 OLI multispectral image and the CBERS-04 P5 image.The GS,PCA,MLT,and Brovey methods were compared in their adaptation to deal with the fusion of the image of Wuzhong City,Ningxia Hui Autonomous Region.The fusion effects between the WT-MB and other four methods were also compared in terms of quantitative evaluation,visual interpretation evaluation and SVM classification accuracy evaluation.The classification accuracy of fused images was judged by the overall accuracy and Kappa coefficient.In addition to enhancing the space information of the OLI ima-ge,the WT-MB transform fusion method could maintain the spectral information of the OLI image more effectively,improve the accuracy of image classification and serve the environmental remote sensing monitoring more accurately and effectively.
作者
郭小丹
GUO Xiaodan(College of Information Engineering,Xizang Minzu University,Xianyang 712082,China)
出处
《华南师范大学学报（自然科学版）》
CAS
北大核心
2019年第4期76-85,共10页
Journal of South China Normal University(Natural Science Edition)
基金
国家自然科学基金项目(41361044)
西藏自治区自然科学基金项目(2016ZR-MY-01)
关键词
图像融合
小波变换
高频空间信息
低频光谱信息
SVM分类
image fusion
wavelet transform
high frequency space information
low frequency spectral information
SVM classification
