Research On Infrared And Visible Image Fusion Based On Pulse Coupled Neural Network

The fusion of infrared image and visible image is a representative research direction in the field of image fusion because of the difference of imaging principle and image characteristics between the two images.Pulse coupled neural network Network(PCNN)is a kind of artificial neural network constructed by simulating the visual nerve activity process of mammalian image processing.Because the simulated neurons can fully consider the global characteristics of image information in the process of fusion through the global coupling of linear addition and nonlinear multiplication;because of its good biological background,the fused image conforms to the visual habits of human eyes It has been widely used in the field of infrared and visible image fusion.But PCNN model also has too many parameters,and the processing of detail texture is not detailed enough,which leads to the poor fusion results.Aiming at the shortcomings of the existing PCNN based infrared and visible light fusion algorithm,this paper proposes two new algorithms combined with non subsampled shear wave transform(NSST).The main contents of this paper are as follows:(1)Infrared and visible image fusion algorithm based on improved simplified PCNN and intersecting visual cortex modelIn order to solve the problems of the existing PCNN based infrared and visible image fusion model,such as many parameters,difficult to determine and poor texture fusion effect,this paper proposes an improved algorithm which combines the simplified PCNN model with the intersecting visual cortical model(ICM)to fuse the infrared and visible images in the NSST domain.After NSST decomposition,the low-frequency components are fused according to the edge and gradient energy,the high-frequency components of level 2 and 3 are fused by the improved simplified PCNN model,the high-frequency components of level 4 are fused by ICM,and finally the final fusion image is obtained by NSST inverse transformation.The PCNN model used in this algorithm is simplified to reduce the model parameters;the external excitation of PCNN is improved to better capture the edge details of the image in alldirections.ICM model itself is good at capturing the fine image features,and the whole can enhance the extraction of edge features of the fusion image.The superiority of the algorithm is verified by experiment,and it performs well in both subjective and objective evaluation.(2)Infrared and visible image fusion algorithm based on connection synaptic computing network in NSST domainIn view of the shortcomings of the existing fusion results based on PCNN infrared and visible image fusion algorithm,such as the lack of detail information and the insufficient retention of edge contour information,this paper proposes the connection synaptic calculation network(LSCN)model and the combination of co-occurrence filtering(COF)algorithm for the fusion of infrared and visible images in the NSST domain.LSCN model has a good restraint of pseudo Gibbs phenomenon,and the image detail information is kept intact,which is used to fuse the low-frequency components of the image;COF filter combined with the maximum and minimum filter can extract the contour information of the image to a great extent,and smooth the texture information,which is used to fuse the high-frequency components of the image.In the algorithm,the problem of infrared image and light image can be enhanced by histogram equalization and LSCN model respectively.The experimental results show that the edge between the target and the background of the fused image is clear,and it performs well in both subjective evaluation and guest evaluation.