The Multisensor Information Fusion Technology For Remote Sensing Image

Image fusion is the process of incorporating images obtained from different sensorsinto a new image by a fusion algorithm, such that the new image is more reliable, clearerand more intelligible. In recent years, multi-sensor image fusion methods and algorithmshave been developed quickly with the development of image sensors. Image fusion hasbeen an important and useful technique for remote sensor image analysis andunderstanding. From the abstract level of the data which is to be deal with, the imagefusion technique can be divided into pixel level, feature level and decision level. Theimage fusion technique in this paper is based on pixel level, and a comprehensive anddeep study on multi-spectral image/panchromatic image fusion is presented. In thispaper, we proposed two novel image fusion methods as follows:(1) Image fusion based on PCA and wavelet packet decompositionFirstly, because of the drawbacks of sensors, people need the help from image fusionmethod to get the remote sensing images, this method can provide good fused image.Secondly, it overcome the drawbacks of traditional fusion techniques, and can avoid thedrawbacks based on the comprehensive analysis on such reasons of the drawbacks. And,it use the wavelet packet decomposition to analysis the images more comprehensive, andpreserve the high frequency information of the image most.(2) Region based image fusion using SVMFirstly, according to the pixel level fusion method, it converts the fusion probleminto the classification problem, it views the image fusion from another angle. Secondly,because of the image characters and content of the remote sensing image, it segmentedthe remote sensing images, and fused the images according to different regions, it avoidsthe drawbacks of the single fusion rule. And, it obtained the fusion rule by SVM to avoidthe drawbacks by using wrong fusion rule. This method is more precise and moreefficient.Based on above, this paper firstly introduce the traditional fusion method, then givemore explanation .of the two fusion method, and provide the experiment result and data tovalidate the analysis in this paper, at last we give the conclusion.