Research On Bistatic Seabed Reverberation Image Fusion Method Based On Graph Structure And Phase Congruency

With the continuous exploration of the ocean by human beings,acoustic seabed imaging technology has become a research hotspot.The monostatic sonar imaging system uses backscattering to obtain the scattering intensity of the seabed but its operating range is short.The bistatic imaging system uses separate vertical transmitting array and towed horizontal receiving array for imaging,which can make full use of the bottom scattering signals from all directions.During the movement of the towed horizontal receiving array,several images that carrying the seabed landform information will be obtained.It is of great significance to study the fusion of the several seabed images.However,the researches of bistatic seabed reverberation image fusion is relatively less.This paper is devoted to studying a fusion method for bistatic seabed reverberation images which can enhance the structure and details of the fused image.The bistatic seabed reverberation image is obtained by homing the beamforming result of towed line array.In seabed images,the direct wave and multipath of tow ship noise are shown as several ray-like echoes with high-intensity which seriously affects the quality of images.Based on inverse beamforming(IBF)interference canceling algorithm,the time domain inverse beamforming(TIBF)interference canceling method is proposed in this paper.The local reverberation level is used to detect the time when interference and reverberation are aliasing.At these times,the beamforming output will be cancelled by the average power of tow ship interference.TIBF method has been verified by simulation in this paper that it can cancel the interference and retain the seabed reverberation information as much as possible.In this paper,the signals on reverberation image are modeled as a gaussian markov random field(GMRF),and the graph structure estimation is transformed into the maximum a posteriori estimation of the GMRF accuracy matrix.The block coordinate descent algorithm is used to estimate the accuracy matrix iteratively.The simulation results show that compared with graph topology inference(GTI)method,the error of graph structure estimated by GMRF is smaller and is closer to the real graph structure.Besides,the estimation algorithm based on GMRF has faster operation speed and lower time complexity.In this paper,the structure of reverberation images and the information redundancy between images are fully considered.Seabed images are firstly decomposed by nonsubsampled contourlet transform(NSCT)to form multiple sub-images of low frequency and high frequency,which are in different directions.And then the multiple sub-images are fused by the estimated graph structure and phase congruency.Finally,the fused reverberation image can be obtained by the inverse NSCT transform.By processing the actual data of the South China Sea acoustic imaging experiment in August 2020,this paper verifies the feasibility of the TIBF algorithm.After interference canceling,this paper fuses the reverberation images,and compares them with other fusion methods.The results show that the image fusion method based on graph structure and phase congruency retains more edge information and structure information of the source sonar images,and can better highlight the strong features in the source sonar images.