Multiresolution Image Fusion And Its Realization

Image fusion is to synthesize source images from different kinds of sensors or the same sensor in different times of a same scene. The fused image can offer a better description and reduce the uncertainty of the scene. With the development of image fusion technology, its applications have branched into medical imaging and diagnosis, remote sensing, intelligent traffic, military and so on.This paper thoroughly researches the multiresolution image fusion algorithms and its realization. Fristly, different pyramid transforms based image fusuion methods are realized and an objective evaluation is given later. Due to the disadvertanges of Mallat and a trous wavelet, Contourlet is introduced as a complement. Contourlet is shift-variant, which will cause artfacts in the fused image. So NonSubsampled Contourlet Transform (NSCT) is used to avoid Contourlet's shortcomings.Pulse-Coupled Neural Networks is shift-invariant, scale-invariant and shift-invariant, the characteristic of which makes PCNN a superb image processing tool. PCNN can synthesize local important information well. This characteristic makes it suit for image fusion. A PCNN and multiresolution decompose method is proposed, which combines the advantages of PCNN and multiresolution transfrom and leads to a better fused image.A high resolution plant cell multifocus image fusion system is realized based on OMAP3530 offered by Texas Instruments, which solves the limited depth of field of the digital microscope at the end of this paper.