Automatic Detecting Outliers in Multibeam Echo Sounding Data

Abstract

Advances in technology and more exacting requirements for hydrographic survey have led to greatly increasing data rates and densities for new generation multibeam systems. It needs rapidly to process the large datasets. However, interactive plot is employed to edit data manually in some hydrographic software. The efficiency is too low. Therefore, some automatic and rapid approaches must be developed. In this paper, an automatic algorithm for detecting outliers is proposed. The algorithm avoids the slow and long time in manual editing. The interactively editing theory is used to automatically detect large and collective outliers, so it is robust. Erosion and dilation effectively retain the normal data which is separated from main block. The result of median filter is not ultimate result, and it is only used to calculate local variance, so distortion in details is avoided and the little outliers are located. It is one kind of postprocessing method. The algorithm is verified by some data and proved to be rapid, robust and able to automatically process large quantities of data.