Research progress on automated detection and cleaning of abnormal ocean multibeam bathymetry data

Aiming at the automated detection and cleaning of abnormal ocean multibeam bathymetric data, and integrating the research progress at home and abroad, we classified the methods into three categories according to the different processing objectives: detection and processing in bathymetric point data domain, in single ping data domain, and in the domain of constructing surface model. By sorting out the median filtering, clustering algorithm, CSF algorithm, CUBE algorithm, trend surface method, and robust estimation method, we summarized the differences in the processing process, application targets, application criteria, application fields, and result judgment of the methods; classified and compared the three types of methods in a list; and obtained the direction of focus and the type of abnormal data applicable to the processing of the three types of methods. The advantages and shortcomings of the three types of the methods for different targets were analyzed, and the problems and the solutions in the processing and practice of the previous methods were reviewed.