Research On The Method Of Multi-source Remote Sensing Image Fusion Based On EMD

With the rapid development of remote sensing technology,the fusion of multi-source remote sensing images has become a research focus in the fields of aviation and satellite image processing.Remote sensing image fusion can not only improve the availability of image information,integrate the complementary information of source images,reduce the difference of single remote sensing images,and obtain richer and more accurate fusion images,and the amount of information after fusion is rich and clear Enhanced,with a higher ability to interpret.This paper mainly studies the fusion method of multi spectral image and panchromatic image of the same ground object.Based on the traditional empirical mode decomposition(EMD),an interpolation method based on particle swarm optimization radial basis function neural network weight is proposed.Based on this,a method of remote sensing digital image fusion based on color space(hue,saturation,IHS)transform,wavelet filtering and improved two-dimensional EMD is proposed.The fusion method can not only retain the spectral information of the fusion image,but also improve the spatial texture details.The main research work and innovation of this paper are as follows:1.This paper describes the research background and significance of remote sensing digital image fusion algorithm as well as the research status at home and abroad,discusses the fusion methods,fusion rules,fusion result evaluation standards in the process of remote sensing digital image fusion,and finally analyzes and compares several commonly used image filtering methods.2.The basic principles and experimental procedures of one-dimensional EMD and two-dimensional EMD are introduced respectively,and the problems needing attention in the decomposition process of two-dimensional EMD are pointed out.3.In the process of image decomposition,the two-dimensional EMD method is improved.According to the interpolation fitting method of envelope sur-face in the process of two-dimensional EMD decomposition,the weights of radial basis function neural network are optimized based on particle swarm optimization.It is verified that the interpolation method based on Particle Swarm Optimization of radial basis function neural network weights has the advantages of high interpolation accuracy compared with radial basis function interpolation method.Finally,when using the improved two-dimensional EMD method to decompose remote sensing digital images The image can be fully decomposed.4.For remote sensing digital image fusion,this paper proposes an image fusion method based on IHS transform,wavelet filter and improved two-dimensional EMD method,and carries out four groups of simulation experiments.The experimental results show that the fusion image can not only retain the spectral information to the maximum extent,but also improve the spatial texture details compared with the original image.Finally,by analyzing and comparing the four experimental results,as well as the panchromatic background image and high-frequency information surface after wavelet filtering,the application scope of the algorithm in this paper is obtained,that is,the algorithm in this paper is suitable for the remote sensing digital image fusion with rich and distinct ground feature information.