Research On Pixel-Level Multi-Source Image Registration And Image Fusion

Multi-sensor image fusion is the technique that integrates complementary and redundant information of multi-sensor images to obtain a composite image, which contains more accurate description of the scene, for observation or subsequent processing. It has been widely used in many fields. However, there are still lots of theoretical and technical problems needed to be solved.In this dissertation, in-depth research has been done on pixel-level multi-sourse image fusion. The main contributions of my jobs are summarized as follows:1. Multi-sourse image registration as the pre-processing process of multi-sourse image fusion has been studied in this dissertation. A multiresolution pyramid transform and joint histogram based fast image registration method is proposed for the registration of multi-source images obtained by similar sensors. A conter-based two step multi-source image registration algorithm is proposed for the registration of multi-source images obtained by dissimilar sensors.2. The color multifocus image fusion algorithm has been studied in this dissertation. A color multifocus image fusion based on principal component analysis and discrete-wavelet frame tansform method is proposed.3. Foused on dissimilar sensors images fusion, three classes of fusion methods are developed: the region-based multi-sensor image fusion algorithm, the multi-source color image fusion nethod, and the HSI color space based false color image fusion algorithm. The three dissimilar sensors images fusion methods details are listed in this dissertation.