The Applications Of Multi-scale Geometric Analysis In Remote Sensing Image Fusion

Remote sensing image fusion is a process that multiple images from the same scene are processed to get a new image. The new image improves the visual effect and extracts the features of original image, which provide richer, more useful and reliable information for practical applications.For the shortcomings of the traditional image fusion algorithms and the merits of multi-scale geometric analysis, the multi-scale geometric analysis Contourlet transform image fusion methods are mainly studied in this paper, as well as some of the factors affecting the quality of fused images. The main contents of this article can be summarized as follows:Firstly, the multi-scale geometric analysis methods and its strengths and weaknesses of applications in remote sensing image fusion are introduced, and the multi-resolution analysis of Piella general fusion model, which is the basis for this fusion algorithm theoretical framework, is also introduced.Secondly, the theory and the nature of Contourlet transformation are analyzed and a Contourlet transform-based remote sensing image fusion method is proposed. Contourlet transform is better able to extract the source image edge details and provide more effective information for the fused image. The experimental results show that the proposed fusion method improves the image spatial resolution and maintains the spectral information of the source image, and the fused images' quality is good.Thirdly, the principles and characteristics of the 1αβcolor space are analyzed, according to the merits of lap color space and Contourlet transform, a fusion method is proposed which is based on the combination of lap color space and Contourlet transform. After the original image is decomposed, the local area of energy strategy is adopted when fusing the high-frequency coefficients. Experiments show that this method not only improves the spatial details of the source image, but also maintains a good image of the spectral characteristics, superior to the traditional fusion method.Finally, when the multi-scale geometric analysis Contourlet fusion method is applied, the effects of different decomposition levels and fusion strategies on remote sensing image fusion quality are studied. The study results show that different levels of multi-scale decomposition have serious impact on fused image quality; with the increase of decomposition level, the spatial details become strengthened, but the spectral information distortions occurs. The contribution of multi-spectral image decomposition high-frequency coefficients to the fused image is different, and it is this distinction has led to the difference of fused image.