Backscatter Data Processing And Image Mosaic Of Multibeam Bathymetry Sonar

The abundance of marine resources has made people realize the importance of surveying and developing the ocean.Compared with the extremely inefficient manual sampling method,the current survey methods have made great progress.Multi-beam measurement system is widely used because of its high accuracy and high resolution,so how to obtain high-quality sonar images is of practical significance.To this end,this paper conducts technical and experimental research on the mapping of seabed backscattering data of multi-beam sonar and the feature matching and mosaic problems in the process of multi-beam sonar image stitching of adjacent routes.First,carry out research on multi-beam sonar backscattering data correction and imaging methods.Based on the angle and time information measured by the multi-beam sonar,combined with the sound velocity profile information and the attitude sensor data for sound ray tracking and spatial homing calculations,the spatial position of each beam is obtained.In terms of echo intensity information,this article combines propagation loss and seabed terrain factors to modify it to obtain backscatter intensity data.In order to eliminate the close connection between the backscattered intensity data and the incident angle and make the backscattered image reflect the real seabed topography,this paper compares the angle response correction of the backscattered intensity data from the three aspects of scattering model,empirical correction model and unsupervised classification method.In addition,considering the seabed backscatter imaging,there is often a band noise phenomenon along the route direction.The noise elimination methods are compared from three aspects: direction filtering,mean normalization and moment matching.Secondly,the research on the feature point matching of the backscatter image of adjacent tracks is carried out.Taking into account the positioning deviation of the backscatter images of adjacent tracks,in order to obtain the precise relative position relationship between adjacent routes.This paper starts from the feature point matching,and combines the experimental data to compare three feature point matching algorithms including SIFT,KAZE and based on neural network models.Finally,the research on mosaic technology of multi-beam backscattered images is carried out.Taking into account the existence of local positioning errors in adjacent tracks.In order to obtain a clear multi-beam sonar seabed backscatter image,this article starts from the local deformation of the image,and compares the two methods based on grid optimization and deformation field to eliminate the local image distortion.The methods realize the shape preservation of the target in the overlapping area after inlay.