A Study Of Optimal Bands Selection And Evaluation Methods On Performance Of The Multi-Source Remote Sensing Image Fusion

Remote sensing fusion is a technology which integrates multi-sensor remote sensing images. In order to enhance the transparence of information; improve the precision, reliability and efficiency; and describe the target congruously it is used to synthesize the band information offered by single sensor or different sensors. It is one of the effective approaches to resolve data display of multi-source images by image fusion.There are a lot of fusion methods which are used in research work. Different fusion results can be obtained by different ways. How to select the best bands combination and evaluate the quality of fused image is an important step of image fusion. Currently, it is common that much attention pays to information fusion research than quality evaluation of fusion results. As band selection affects the fusion results directly while the fusion results reflect the advantage and disadvantage of the methods, it is necessary to study on the best band combination and the quality evaluation of fused image.On the background the thesis focuses on the band selection and fusion image quality evaluation according to the spectrum characteristic of sensors, spectrum radiation energy of object and correlation of each band. The main works of the thesis are as following: According to the spectrum of each band and evaluation criteria of SPOT5,QUICKBIRD,IKONOS,IRS-P6, we can find the best band combination through the statistic of data quantity and correlation. The four kinds of data are divided into two catalogs by band characteristic. The best fusion method can be got according to evaluation of color and resolution between fused images and original images. Finally, we obtained the best band combination and the best fusion method of the two catalogs of data.The research conclusion can be used in the researches about image fusion.