Research On Image Fusion Algorithms Based On PDE

In recent years,edge-preserving filters have been widely used in the field of image processing and has obtained the good research result.As a multi-scale decomposition tool,the filter can decompose the source image into a smooth base layer and one or more detail layers of different scales.The partial differential equation(PDE)is transformed into a nonlinear iterative filter after effective discretization,which can accurately separate the fine scale texture details,mesoscale edges and large scale spatial structures of the image.This feature helps to reduce halo and aliasing artifacts in image fusion.Therefore,the partial differential equation is used as the edge-preserving filter decomposition tool.It is of great theoretical significance and practical value to study the image fusion algorithm based on PDE.The specific research content and relevant conclusions of this paper are as follows:(1)This paper briefly introduces the classical second-order P-M model and fourth-order Y-K model in image denoising,and proposes a new nonlinear PDE model combining the second-order and fourth-order.Firstly,a new diffusion function is constructed,and its smoothing effect is better than the diffusion function in P-M model and proposed by Tebini et al.Secondly,a new PDE model is proposed by introducing a new weight function,which overcomes the "ladder effect" and "speckle effect".The numerical experiments show that the proposed model has good ability of denoising and edge preservation.(2)The proposed non-linear PDE model is introduced into image fusion as an edge-preserving filter.The proposed model decomposes the source image into basic layer and detail layer,meanwhile different fusion rules are designed for different layers.Finally,the fused image is reconstructed linearly through the detail layer and the basic layer.The performance of the fusion algorithm is evaluated by simulation experiments and further compared with the classical edge preserving decomposition algorithm.The results show that the proposed image fusion algorithm based on nonlinear PDE has better fusion performance for multi-source images.(3)The total variation and P-Laplace image restoration model proposed by Li Dan et al.are improved,and the edge retention ability of the algorithm is improved.In order to make full use of the relevant details of the source image,a new fusion method based on infrared and visible images is proposed on the basis of the PDE restoration model.We uses Gaussian filtering and PDE restoration model to decompose the fused source image into different scales and obtains a basic layer and a series of detail layers of different scales.We design different fusion rules of basic layer and detail layer.The experimental results show that the IR and VIS image fusion method based on rolling PDE maintains the spatial consistency.It has obvious improvement in IR target extraction,VIS background display,visual sharpness and contrast.