Research On Digital Image Fusion Based On Wavelet Transform

With the development of modern remote sensing technology, more and moremulti-resolution, multi-temporal and multi-spectral remote sensing image data areprovided by different kinds of remote sensors. To get a more correct, clear and completeknowing about the target, it is urgent for us to find a way to synthesize different kinds ofimage data. So the technology of image fusion appeared. Image fusion is an integrationof multiple images obtained from one sensor or multiple sensors. In the integratingimage many information for multiple original images are contained to analyze andjudge the objects more accurately and perfectly. Image fusion is a new research field, itis widely applied in many fields, such as computer vision, remote sensing, militaryobject recognition and medicine image.From80’s up to today, researchers have put forward many different algorithmmethods for the various objects, of which the most representative are those based onpyramids. But these methods are subjected to some restrictions in their application,each one can only be applied in one field. Wavelet transfer is a time-frequency anslysismethod that has developed in the last twenty years.It is a powerful tool to represent thesignals in different levels and resolutions.Now wavelet is wildly used in the fields ofsignal processing,image processing,pattern recognition,machine vision and so on.In this thesis, in the first and second chapter a brief image fusion depiction is given,including the history context, present situation and development of the image fusionresearch.In the chapter2,the level,steps and methods of image fusion are introduced.Inthe method of image fusion,the method based on spatial domain and transform domainare proposed.At the end of this chapter,the quality evaluation of image fusion isintroduced.In the third chapter,simple method of image fusion is given,such as selectedweighted verage pixel,small pixel value or large pixel.Thesis compare a variety ofmethods of image fusion by simulation experiment. The chapter4is mainly based on pyramid decomposition of image fuion method.This chapter focuses on ths imagefuson based on Laplacian pyramid decomposition.The rule based on pixel and rulebased on regional character are given in the rule of this method.In the end of thischapter,thesis analysis the effect of image fusion by simulation experiment.The chapter5is the core chapter of the topic.We first decsriebe the theory of wavelettransform,multiresolution anslysis,and the Mallat algorithms for two-dimensionlimage decomposition and construction.And then we briefly introduced the step ofimage fusion based on wavelet transform.Next,the choice of the decomposition leveland the fusion ruler for the Wavelet based on image fusion method are discussed.Thispaper introduces several fusion rules about high-frequency,low-frequency waveletcoefficient. We analysis and compare the effect of image fusion based on differentfusion rule through simulation experiment.The last chapter is image fusion based onwavelet transform and color space transform.This chapter mainly discuss the commoncolor space model and the convertion between each other. The method and step aboutimage fusion based on wavelet transform and color space transform are proposed.Themethod is verified through simulation experiment.