The Research Of Multiphysics Image Processing Method Based On PDE

In the late twentieth century,many scholars began to apply partial differential equation(PDE)to image processing,and achieved many meaningful results.In recent years,PDE image processing mainly focuses on the research of non-linear diffusion methods of pixel gray level guided by image gradient in image spatial domain.These methods can remove noise and keep image edge details well.However,the regularization model of PDE needs to pre-filter the image according to the degree of image noise pollution,and can not realize real-time image processing.In order to remedy this shortcoming,a modified multi-physical field coupled PDE model processing method is proposed based on the existing image processing methods for the interferometric synthetic aperture sonar(INSAS)image in physics.The specific work is as follows:Firstly,the common denoising methods of In SAS interferograms are studied.The median denoising method,mean denoising method and wavelet denoising method commonly used in digital image denoising are described.By researching these three common methods and using MATLAB software,the three common denoising methods are numerically implemented for general images and in-SAS interferograms.The denoising effects of these three denoising methods for general images with less noise and in-SAS interferograms are analyzed and compared.Four evaluation methods of image denoising quality are given.Secondly,the variational method of image processing and the PDE method are compared.The basic theory and principle of variational method are expounded,including how to transform it into Euler equation and how to find the minimum extremum of Euler equation's functional.At the same time,the image processing methods of several existing PDE models are studied,including the TV model of image denoising,the linear diffusion model of image denoising,the P-M model of image denoising and Chen's PDE coupled denoising model.With the help of COMSOL Multiphysics software,these PDE denoising methods are used to solve the interferogram numerically,and the effect of image denoising is analyzed according to the given image quality evaluation method.The results show that the linear model has better ability of denoising and edge preservation,the TV model has the strongest ability of denoising,and the P-M model has the strongest ability of edge preservation.Finally,the modified PDE model of multi-physical field coupling denoising is established and studied.Aiming at the problem of image processing of in SAS interferogram,COMSOL Multiphysics 5.3a software is used to simulate the seepage of the interferogram with noise.By comparing the seepage field with different physical parameters and different diffusion effects,the appropriate seepage field image is selected,and the geometric structure of the original image is adjusted with the obtained seepage field,so the guided non-linear expansion can be obtained.The whole process of dispersal.In particular,the results of the coupled denoising model are compared and analyzed,and the multi-physical field coupled PDE model is optimized.A velocity term is added to the diffusion function to smooth the diffusion of the original model.Based on this,a modified model of the multi-physical field coupled model is proposed and established.With the aid of numerical simulation,the PDE model image processing method is used to solve the interferogram numerically,and the denoising effect before and after model modification is verified.The results show that the modified model has better edge preservation ability and avoids the pre-filtering of the regularized model.