The Research On Multi-sensor Image Fusion Algorithm

With the rapid development of the sensor technology,more information is obtained.The single sensor cannot effectively deal with these information,so the multi-sensor image fusion technology began to develop.Image fusion can comprehensively deal with useful information of images from multiple sensors to obtain a new image which contains more abundant?more accurate information.In recent years,many researchers have joined in the team of exploring the methods of image fusion.Because the the image fusion theory become more rich,its application also gradually become stronger.This paper simply introduces the research background of multi-sensor image fusion.And some commonly used image fusion methods are reviewed in detail.The subjective and objective evaluation methods are also introduced in this paper.According to different research purposes,several new fusion algorithms are proposed in this paper.And they can obtain good fusion results.The main work of the paper is as follows:1.Considering the imaging mechanism of infrared and visible images,it is proposed to a new fusion algorithm of infrared and visible images based on NSCT.In order to fully extract the target characteristics of the infrared image and the background information of the visible image,the improved fusion rules are proposed in the low frequency part and the high frequency part.The fusion rule based on comparing the subtraction of normalized regional variance with the threshold is used in the low frequency coefficients.And the fusion rule based on combining region energy with sum of Laplace energy is used in the high frequency coefficients.The contrast experiments show the proposed method has better result than traditional image fusion methods.It can more accurately reflect the research scene.2.Because PCNN can effectively extract the detail information of the image,the PCNN model is used in the high frequency part.Considering the advantages of NSCT,such as translational invariance,it is proposed to a novel image fusion method combing the nonsubsampled Contourlet Transform(NSCT)with pulse coupled neural networks(PCNN)in the paper.The fusion rule combing the improved edge energy with the spatial frequency is used in the low frequency part.The simplified mathematical model of pulse coupled neural networks(PCNN)is used in the high frequency part.The improved sum of Laplace energy is used as the linking strength of pulse coupled neural networks(PCNN).And it is selected to the method of combing the fire numbers with the standard deviation of the fire mapping images in the high frequency part.The analysis of the experimental results can fully prove the validity of the new method.3.Because the IHS model is closer to the human eye visual feature than the RGB model,the remote image is converted into the IHS space from the RGB space.In addition,IHS transform can separate the I component?the H component and the S component.So it is helpful to retain the spatial information and the spectral information.It is proposed to an improved remote image fusion method combing NSCT with IHS in the paper.The adaptive fuzzy logic method is used in the low frequency part.The method of the improved energy maximum is used in the high frequency part.The experimental results show that the fusion image of the new method can reserve more complete scene information.