Research On PanSharpening Fusion Method Of Remote Sensing Images

With the development of satellite remote sensing technology,there are more and more types of remote sensing images available,and the information contained in remote sensing images tends to be diversified and complicated.For various needs in various fields,there are many classifications of remote sensing image processing.Among them,the fusion of remote sensing images is one of the important directions.Image fusion is to select and fuse images from different imaging sources into an image that satisfies requirements.The fusion of multispectral images and panchromatic images is very widely used.Due to the constraints of the imaging principle of satellite sensors,we often only get low-resolution multispectral images and high-resolution panchromatic images.The research content of this paper is the fusion of multispectral images and panchromatic images.Multispectral image fusion,also known as PanSharpening,The goal of PanSharpening is to increase the spatial resolution of the multi-or hyper-spectral images using a high spatial resolution panchromatic image.PanSharpening methods are mainly divided into component substitution-based methods,multiresolution analysis-based methods,and model-based methods.Traditional image fusion methods based on component substitution include IHS method,PCA method,GS method,etc.;Image fusion methods based on multi-resolution analysis include wavelet transform,Laplacian pyramid decomposition,and CT transform method.Traditional methods have simple framework,and easy to implement.However,there are serious spectral distortions.Avoiding the spectral distortion of fused images is the primary task of fusion research.This paper proposes a method that is non-linear IHS method combined with NSCT method.The nonlinear IHS is an improvement over the traditional IHS method.This method approximates upsampled intensity by segmenting the upsampled multispectral image and applying a linear model.The upsampled intensity component can deal with the local dissimilarities and describe the spatial structure of the upsampled multispectral image more accurately,thereby reducing the problem of spectral distortion after fusion,and the nonlinear IHS method is based on the IHS method,the framework is simple and easy to implement.The NSCT translation of the upsampled intensity component and panchromatic image is performed to obtain the high frequency coefficients and low frequency coefficients.The high frequency coefficients and low frequency coefficients are respectively fused using the weighting method and the edge intensity rule,and then the new coefficients obtained by inverse NSCT.After the fusion of the upsampled intensity components,the fused image is finally obtained using the generalized IHS transform formula.Through the combination of the two methods,the down-sampling operation of the image is avoided,so that the fusion image can better maintain the spectral information of the original multi-spectral image while the spatial details information is injected.In this paper,we use the GF-2 satellite remote sensing data to verify the method of this paper,and compare the experiment with the traditional method.The experimental results show that the proposed method has a better fusion effect in terms of objective quantitative and subjective perception.