Research On Image Fusion Based On Wavelet Transform

Image fusion means using fusion algorithm for two or more images from multiple sensors, to obtain a clearer description of an object. It has great applicational value in medical imaging, military areas, target detection systems, remote sensing images and other fields, prompting a lot of researchers to make in-depth researches.This thesis introduces the background information and significance of image fusion, and its different levels including pixel-level image fusion, feature-level image fusion, decision level fusion. The introduction focuses on the level of pixel-level image fusion algorithms based on the maximum fusion method, the weighted average method, PCA, IHS transform, and wavelet transform. In wavelet theory, formulas are described in detail, including the basic concepts of wavelet transform, the wavelet function created by wavelet generating functions, the definition of continuous wavelet transform, the definition of discrete wavelet transform, and the definition of two-dimensional wavelet transform. Based on the wavelet theory, the basic wavelet transform image fusion process is described, and the theoretical knowledge of the Fractional Wavelet transform and its implementation is studied.Based on the study of Fractional Fourier transform, one of the definition of the Fractional Wavelet transform is explained.In the field of signal processing, a function, the result of its Fourier transform being equal to its own, is called a Self-Fourier function; in the same way, a function, the result of its fractional Fourier transform being equal to its own function, is called a Self-fractional-Fourier function. These functions have the properties of symmetry, displacement, convolution, etc. In this thesis, the function definition and properties are described in detail. Together with Fractional Wavelet transform and Self-fractional Fourier function, a new image fusion algorithm named SFFF-FRWT is presented. Through a series of rigorous experiments, the optimal parameters of the algorithm are obtained, and simultaneously, the analysis of the comparative result with other image fusion algorithm, verifies that the SFFF-FRWT image fusion algorithm has more feasibility and advantages.According to the second definition of Fractional Wavelet transform,which struct a function of fractional wavelet function as the wavelet basis function to form the Fractional Wavelet transform. In this thesis, the definition of this form of fractional B-spline wavelets body is investigated, and the IHS, PCA transform image fusion algorithm theory and processes are discussed. Then together with the symmetric Fractional B spline wavelets, IHS, PCA, a new image fusion algorithm is presented. Through rigorous comparative experiments, I compare this algorithm with other image fusion algorithm, so as to analyize the low-resolution multi-spectral images and panchromatic images. According to my subjective and objective judgment, the proposed algorithm is able to guarantee the spectral characteristics and high-resolution remote sensing image characteristics.