Research Of Image Fusion Algorithm Based On Satellite Cloud Image

Employing data of infrared cloud image and visible cloud image providedby meteorological satellite, this dissertation is aimed at improving resolutionratio of infrared cloud image by multiscale analysis method and lay thefoundation for veracity of detection and early warning of Strong convectiveclouds.This dissertation focuses on Multi Waveletlet Transform and ContourletTransform and their application in Image Fusion, implementing several aspectsof performance as following:Discuss general theory of Waveletlet Transform and Multi WaveletletTransform and image binary Waveletlet Transform algorithm.Study ImageFusion algorithm on the basis of Multi Waveletlet Transform.In consideringphysical significance of satellite cloud picture,fuse Texture details information ofvisible light holding high spatial resolution into infrared cloud image holdinglow spatial resolution on Visual C++platform. Experiments prove that thismethod produce infrared cloud image of high spatial resolution and enhancequantity and accuracy of Physical information of infrared cloud image.This dissertation intensively analyze coalescing NSCT(NonsubsampleContourlet Translation) and PCNN(pulse coupled neural network) and raise afresh algorithm appropriate for image fusion of infrared radar image and visiblecloud image. Utilizing multiscale and multiaspect and translation invariance ofNSCT, put forward new fusion rule of PCNN. This algorithm improves problemof traditonal methods which may lose partial useful information of image andcontain more information of source image. The method also increase recognitionrate of objection, making infrared cloud image holding more texture informationthan before, thus achieving better forecast of formation of convective cloud.