Research On Thermal Infrared And Visible Images Fusion

The fusion of thermal infrared and visible images has been the focus of image fusion research,and it has wide demand and application in military detection,security monitoring and other fields.This paper is based on the infrared and visible image fusion frame model based on the adaptive PCNN and information extraction in the NSCT domain,with the thermal infrared image and the visible light image as the research object.The main contents of this paper are as follows:(1)Four algorithms of Mallat,Curvelet,Contourlet and NSCT commonly used in multi-resolution analysis are introduced.The correlation between these four algorithms in multi-scale decomposition and reconstruction of images is compared,and the fusion of infrared and visible images by multiresolution analysis is also studied.The results show that the four algorithms can all decompose and reconstruct better images.The fusion results of infrared and visible images show that the fusion of infrared and visible light images based on the NSCT algorithm is the best.On this basis,the infrared and visible light maps based on the adaptive PCNN in the NSCT domain are further studied.Like fusion.(2)In view of the characteristics of infrared images,a fusion algorithm of infrared and visible images based on adaptive PCNN based on the combination of pulse coupled neural network and Otsu method is proposed.The experimental results show that the adaptive PCNN algorithm can effectively extract the target features of the infrared image,the fusion image quality is high,the fusion image inherits the characteristics of the target clear in the infrared image,and also inherits the details of the edge and texture in the visible light image.(3)On the basis of the fusion of infrared and visible images based on adaptive PCNN,a fusion algorithm of two infrared and visible images based on adaptive PCNN and information extraction is proposed based on the adaptive PCNN and information extraction in the NSCT domain.The fusion method of the first fusion image and the original image is formulated by objectively evaluating the information entropy of the index and the edge retention.The experimental results show that after two fusion images,the amount of information has been significantly improved compared with the two original images.Compared with the first fusion image,the target area of the second fusion image is obvious,the character image is remarkable,the target features are obvious,the edge contour is clear,and the detail texture features are prominent.The method described in this paper is superior to several popular image fusion methods based on multi-scale transformation from subjective visual effect and objective evaluation.Compared with NSCT based image fusion,cross entropy,standard deviation,peak signal to noise ratio and mutual information are selected,and the fusion images are analyzed objectively by 4 indexes.The two groups of experimental fusion quality indexes are respectively analyzed.