摘要
图像融合的目的是从包含同一场景的互补信息的多个输入图像中生成合成图像,该合成图像比任何单一的输入更适合于人或机器的感知。由于这一优势,图像融合技术在依赖同一场景的两幅或多幅图像的各种应用中显示出重要的意义。随着国内卫星技术的不断发展,各种新型传感器应运而生,不同传感器获得的数据为科学研究提供了宝贵的资源。本文采用基于像素的融合方法GS变换、PCA变换、CN变换对不同数据源的数据进行融合试验,从主观定性评价和客观定量评价两个方面对比分析融合算法的优劣。结果表明,针对GF-1和ZY-3影像,CN变换法是最适合的融合算法;针对GF-2影像,PCA变换法是最适合的融合算法。
The goal of image fusion is to generate,from multiple input images containing complementary information of the same scene,a composite image that is more suitable for human or machine perception than any single input.Due to this advantage,image fusion techniques show great significance in various applications that rely on two or more images of the same scene.With the continuous development of domestic satellite technology,various new sensors have emerged,and the data obtained by different sensors have provided valuable resources for scientific research.In this paper,pixel-based fusion methods GS transformation,PCA transformation and CN transformation are used to perform fusion experiments on data from different data sources,and the advantages and disadvantages of fusion algorithms are compared and analyzed from two aspects:subjective qualitative evaluation and objective quantitative evaluation.The results show that CN transformation is the most suitable fusion algorithm for GF-1 and ZY-3 images;PCA transformation is the most suitable fusion algorithm for GF-2 images.
作者
武萌华
高林营
WU Menghua;GAO Linying(Chongqing Survey Institute,Chongqing 401121,China;Chongqing Engineering Research Center of Geographic National Condition Monitoring,Chongqing 401121,China)
出处
《测绘通报》
CSCD
北大核心
2022年第S02期150-155,共6页
Bulletin of Surveying and Mapping
关键词
国产影像
全色影像
多光谱影像
融合方法
domestic image
panchromatic image
multispectral image
fusion method
