Research On Methods Of SAR And Optical Remote Sensing Image Fusion

With the rapid development of remote sensing technology,it is no longer difficult to obtain a large amount of multi-temporal,high-resolution and multi-polar satellite remote sensing image data.As a widely used remote sensing image,SAR and visible images have different gray values but each have their own image features.SAR images are sensitive to man-made structures through active microwave imaging,and reflect texture features and structural information of objects.Visible image reflected by light,reflecting the spectral information of the image and the general outline of the object,with superior visual effects.However,due to the limitation of the sensor design,the spatial resolution and the spectral resolution of the remote sensing image are balanced with each other.With the increasing demand for remote sensing image data,how to achieve the advantages of SAR images and visible images,and obtain high spatial resolution and hyperspectral image are the research focuses in the field of remote sensing today.In this thesis,starting from the multi-source remote sensing sensor,after analyzing and summarizing the imaging principle of SAR and visible images and the existing fusion algorithms at home and abroad,it is found that the current fusion algorithm still has spectral distortion and the target of interest does not stand out.Based on the existing technology,this paper studies the algorithm of SAR and visible light image fusion.The main work is summarized as follows:A fusion algorithm based on sparse representation of NSST and IHS transformation is proposed.The algorithm takes advantage of the NSST transform in multi-scale analysis and translational invariance,and uses sparse representation to capture structural features.It emphasizes the sparse representation of structural similarity and brightness variability of low-frequency components,enhances the global significant structure and the degree of recognition of the target of interest;The high-frequency components are analyzed for the relationship between the pixels in the local area,and the fusion rules can preserve the details of the image space.A fusion algorithm of SAR and visible combined with sparse representation of Phase consistency is proposed.This algorithm uses the advantage of phase consistency to extract the texture features of images,and maintains the characteristics of phase consistency to smoothly inject the sparse representation.The algorithm takes into account the difference of high and low frequency components,and the targeted design fusion rules make the fusion image be effectively maintained in feature retention,texture edge,and spectral characteristics.The thesis makes subjective and objective evaluation of the proposed algorithm based on the data of landsat8 optical satellite and Sentinel-1 SAR satellite.According to the experiment and comparative analysis,the results show that compared with other fusion algorithms,the proposed algorithm has obviously improved both visual and evaluation indexes,and the spatial structure information and spectral information are effectively maintained.