Research On Infrared And Visual Image Fusion Method

Infrared image and visible light image have different imaging principles,and the information they contained is also not the same.Infrared imaging is based on the infrared radiation of target object,which is sensitive to temperature,but little affected by light.Visible light image is rich in details,but it is greatly influenced by illumination and environment.It can't operate in all weather.Therefore,the fusion of infrared and visible image can combine their advantages to get a complementary image with clear background and prominent targets,which is conducive to observer to make a more accurate and comprehensive description of the scene.Image fused by the traditional image fusion algorithm always contains the problems of lacking in details such as edge,texture information and shortage of background information.Integrated with the latest research achievements inside and outside the country,this paper studies the two problems based on image multi-scale fusion framework.The main research content and innovation of this paper are as follows:(1)The compressive sensing(CS)theory and the linking synaptic computation network(LSCN)model are studied.An infrared and visible image fusion algorithm based on CS,LSCN and improved pulse coupled neural network(IPCNN)is proposed.By combining the static wavelet transform(SWT)with the nonsubsampled contourlet transform(NSCT)algorithm,we can get the image with more sparse expression.According to the distribution characteristics of the decomposition coefficients,CS and IPCNN are applied to the fusion of image components,a hierarchical LSCN enhancement strategy is also proposed.The experimental results show that the fusion image has rich detailed information,and its evaluation criterions are better than several other methods.(2)The spiking cortical model(SCM)is studied.In order to solve the problem that the SCM model is not sensitive to the dark area in image,a dual channel SCM(DCSCM)model is proposed for the first time.This model has fewer parameters than the traditional SCM model and can effectively extract the dark areas in image.In this paper,a novel infrared and visual image fusion algorithm based on improved guided image filtering and dual channel spiking cortical model is proposed.By decomposing the two original images with nonsubsampled shearlet transform(NSST)separately.The improved guided image filtering algorithm and DCSCM model are used to fuse the low frequency and high frequency components respectively.The experiment is carried out with the images to be fused,results show that the fusion image contains well maintained edge and abundant background information,and the proposed method has obvious advantages over several other methods.