Research On Reverse Time Migration Based On High-Order FDTD

With the emergence of digital processing technology,reverse time migration(RTM)has been widely applied in the fields of geophysical exploration,disaster relief,non-destructive testing,and others.Due to the increasing complexity and invisibility of the detection area,traditional RTM imaging algorithms based on electromagnetic wave can result in rough imaging results,or even dislocation of the target position.Therefore,in order to improve the quality and accuracy of imaging,this paper proposes the use of the high-order RTM imaging algorithm under cross-correlation imaging conditions.However,this method can generate a large amount of low-frequency noise between the wave source and the boundary,which needs to be processed.In complex scenes,traditional denoising methods cannot effectively reduce low-frequency noise.To better reduce the interference of noise on imaging,an improved wave field decomposition method combined with singular value decomposition(SVD)denoising method is introduced in this paper.The approach demonstrates superior denoising performance compared to both traditional denoising methods and the single wavefield decomposition method.Specific research work are as follows:(1)Establish an RTM imaging model based on Modified high-order finite difference time domain(MFDTD(2,4))algorithm.In order to address issues of numerical dispersion errors and low computational accuracy in the traditional FDTD,the high-order MFDTD(2,4)algorithm is introduced into RTM algorithm to improve the simulation accuracy of radar echoes in this paper.Meanwhile,high-order Convolutional Perfectly Matched Layer(CPML)absorbing boundary conditions are introduced at the truncated boundary of the computational domain to improve the numerical stability and accuracy of the algorithm.Numerical results demonstrate that the proposed imaging algorithm has higher accuracy compared to the RTM imaging algorithm based on the traditional FDTD method.(2)To address the issue of traditional denoising methods being ineffective in reducing noise in complex scenes,this paper proposes an improved wavefield decomposition method combined with SVD for denoising.This method first decomposes the wavefield using different directional decomposition methods,dividing the wavefield into sub-wavefields with different frequencies and directions,thus removing noise while retaining useful information.Then,SVD is used to perform secondary denoising on the previously denoised imaging results,and the imaging result is reconstructed by retaining the components corresponding to larger singular values.Numerical examples demonstrate that compared with traditional denoising methods or single wavefield decomposition methods,the proposed method has better denoising performance.(3)In this paper,the RTM based on MFDTD(2,4)is used to process simulation data from ground-penetrating radar and through-wall radar,and undulating terrain surface,double-landmine,targets behind walls,and surrounding walls are modeled to verify the applicability of the algorithm in complex detection scenarios.Numerical results indicate that the denoising method proposed in this paper further reduces lowfrequency noise between the source and interface,as well as at large reflection angles,when processing underground models with steep structures.This improves the imaging resolution.In the imaging of wall targets and targets behind walls,the proposed imaging algorithm can more clearly describe the position and outline of the wall and target.