Design And Implementation Of Multispectral Image Fusion System

Image fusion,as an important branch of image processing,refers to the fusion of multi-spectral images acquired by different types of sensors in the same scene into one image,so that it contains most of the data in the multi-spectral image without distortion information.The multi-spectral image fusion technology has been widely used in satellite remote sensing,military and medical fields as soon as it came out.This article will focus on the design and implementation of a multi-spectral image fusion system with perfect functions and good human-computer interaction experience.Finally,based on the fusion system implemented in this paper,an image fusion method based on NSCT is proposed to achieve better fusion results.The research contents of this article are as follows:(1)A detailed overview of the basic theory of image fusion,including the basic definition,hierarchy and workflow of image fusion,based on this,the development environment and tools of the system are analyzed and determined,the core modules of the system are divided,and finally the Open CV machine is applied The visual library and graphical user interface development framework QT designed and implemented the overall framework of the multi-spectral image fusion system.Then introduced the image preprocessing,image fusion and image fusion quality evaluation,and studied representative algorithms.Based on the research of the above algorithm,the basic framework of the algorithm in each core module of the multi-spectral image fusion system is designed.Finally,the Open CV machine vision library is used to encode the algorithm of each core module,and the experiment is displayed to improve the overall function of the software system.And initially verified the feasibility of the system;(2)The image fusion method based on transform domain is one of the research hotspots in the field of image fusion.The key lies in the selection of multi-scale analysis tools and the selection of high and low frequency fusion strategies.First,the multi-scale transform analysis tools and fusion strategies based on the transform domain are introduced in detail.The basic theory of several multi-scale transform tools and their applications in image fusion are introduced,including Laplacian pyramid fusion algorithm and wavelet transform fusion Algorithm and fusion algorithm of nondownsampling contour wave transform.Based on the above research,an image fusion algorithm based on transform domain is designed and implemented,which improves the multi-spectral image fusion system in this paper.Based on the above research,the fusion experiment of the studied algorithm is carried out.Through the analysis and comparison of fusion results,the subjective and objective evaluation is comprehensively considered The non-downsampling silhouette wave transform fusion algorithm has better performance,and the stability of the image fusion system has been verified through multiple sets of fusion experiments;(3)Using non-downsampling contour wave transform(NSCT)as a multi-scale transform method,using an improved adaptive pulse coupled neural network(PCNN)as a high-frequency subband fusion strategy,and using weighted PCA transform as a lowfrequency fusion strategy A multi-spectral image fusion method based on NSCT.Taking the fusion system proposed in this paper as an experimental platform,through the fusion experiment and result analysis of the multi-spectral image fusion system,after comprehensively considering the subjective and objective evaluations,the fusion results of the algorithm proposed in this paper are combined with the Laplace pyramid transform,discrete wavelet transform and non-down Compared with image fusion algorithms such as sampling contour wave transform,the algorithm proposed in this paper has better fusion performance,and further verifies the stability of the image fusion system through multiple sets of fusion experiments.