Research On Remote Sensing Images Fusion Based On Cpmpressed Sensing And Classiifcation Method

Remote sensing images fusion methods based on compressed sensing and remotesensing image classification methods based on spectrum and texture features are studiedin this paper.Remote sensing images fusion methods based on compressed sensing is proposedin this paper. In this method, multi-spectral image is transformed by IHS firstly, then theresult and the full spectral remote image are compressed through measure matrix, andthe compressed images are divided into two components, high and frequency, bytwo-dimensional wavelet transform in the compressed domain. In the high frequency,the high frequency fusion rules with gradient are used, and in the other frequency, thelow frequency fusion rules are took. The fused image after compression is got byinverse wavelet transform with fusion coefficient. Then the final image is got by OMPreconstruction algorithm and inverse IHS transform. The simulation experiment resultsshow that the proposed fusion method compared with the method proposed by JuanjuanHan, at the same time of enhancing spatial detail information, the spectral informationobtained better retention.The study of remote sensing image classification methods in this paper is based onthe research by Zhao Guobin. Based on the GLCM and based on wavelet transformtexture extraction are proposed. And the fuzzy reasoning classifier which takes aminimum distance method as the central is used for classification. Fuzzy membership isgot by fuzzy fusion method combined with spectral features and texture features. Theexperiment which classification with the source image and fused image shows that, thefusion method can improve the image accuracy of classification. The two classificationalgorithms proposed in the paper compare with the basic algorithm, the result showsthat, the separate feature precision and the overall accuracy are improved.