Multi-beam Echo Sounding Terrain Data Processing Method And Visualization

With the development of modern science and technology such as computer technology,electronic technology and echo detection technology,the multi-beam sounding system(MBES)has become the mainstream advanced equipment for marine topography detection with its advantages of full coverage,high resolution and high precision.However,with the development of marine mapping technology,the amount of point cloud data acquired by the field is getting larger and larger,and there is a lot of noise in the data,which brings great challenges to the display and post-processing of point cloud data.Aiming at the above problems,the traditional trend surface filtering algorithm is improved,and an adaptive trend surface filtering method based on K-D tree is proposed;Parallel processing of uncertainty-based multi-beam sounding data filtering algorithm(CUBE)to effectively improve algorithm execution efficiency;An octree-based LOD hierarchical model construction method is proposed to realize the visual display of the filtering processing results.The main research contents and research results are as follows:(1)Multi-beam sounding anomaly data detection and rejection algorithm.The system explains the basic principles,mathematical models and algorithm implementation steps of statistical filtering,trend surface filtering,the robust M estimation filtering and CUBE filtering algorithm.The traditional trend surface filtering algorithm is deeply analyzed,and an adaptive trend surface filtering method is proposed.The design experiment uses the above algorithm to process the measured multi-beam sounding topographic data,and compares algorithm execution efficiency and noise recognition effect.The experimental results show that the PCL statistical filter has the highest execution efficiency,but there is a loss of real information in the edge region;The traditional trend surface filtering efficiency is second only to the PCL filter,and it is unable to effectively identify the near-field noise and preserve the edge region of the point cloud;Adaptive trend surface filtering execution efficiency is about twice as long as trend surface filtering in the case of millions of data volumes.The ability of the algorithm to detect noise and retain true terrain information is superior to traditional trend surface filtering;The robust M filtering algorithm and the CUBE algorithm are consistent in the ability to recognize noise and preserve real terrain,however,the implementation time of the robust M filtering algorithm is about twice that of the CUBE algorithm.(2)Parallel CUBE filtering algorithm.Aiming at the efficiency problem of CUBE,a parallel CUBE filtering algorithm based on K-D(K-Dimensional Tree)tree and GPU acceleration is proposed.The K-D tree is used to index the multi-beam point cloud data.and the GPU's Streaming Multiprocessor(SM)is used to process the CUBE filter calculation in parallel.thereby effectively improving the algorithm...