Universiti Teknologi Malaysia Institutional Repository

Automatic filtering of far outliers in multibeam echo sounding dataset using robust detection algorithms

Mahmud, Mohd. Razali and Mohd. Yusof, Othman (2005) Automatic filtering of far outliers in multibeam echo sounding dataset using robust detection algorithms. In: International Symposium & Exhibition on Geoinformation 2005 Geospatial Solutions for Managing the Borderless World,, 27 - 29 September 2005, Pulau Pinang.

[img] PDF
Restricted to Repository staff only


Official URL: http://www.civil.eng.usm.my/isg2005/home.shtml


Bathymetric data collections using multibeam echo sounder (MBES) have led to increasing data rates and densities. While it is really advantage having full coverage of seabed, data management is the utmost aspect to establish. In this data collection method, part of the dataset contains erroneous data, as measurements are always associated with uncertainties. The critical task for hydrographic surveyor is to make decision on which data can be accepted as good data and the remaining data will be considered as outliers. As there is no ground truth available for the MBES data to compare with, the best solution to address this problem is by using statistical outliers elimination. In order to obtain meaningful results when statistical tools are in used, the dataset should be in a Gaussian distribution. To ensure that the dataset in a bell-shaped curve characteristic, the far outliers must be eliminated prior to any processing. This certainly needs further considerations on characteristics of the erroneous data. Thus, a post-processing program was developed to detect and discard the MBES far outliers based on behaviours of propagated beam in the multibeam sonar system. The entire data have to go through a series of far outliers screening. A remarkable result can be achieved by filtering these far outliers using automatic detection mode. This paper elaborates the techniques used for the detection and elimination of the far outliers in the MBES dataset, known as robust detection algorithms. It also explains on the filtering sequences used and results produced by the developed program.

Item Type:Conference or Workshop Item (Paper)
Uncontrolled Keywords:MBES, far outliers, robust detection
Subjects:T Technology > TA Engineering (General). Civil engineering (General)
Divisions:Geoinformation Science And Engineering
ID Code:1376
Deposited By: Tajul Ariffin Musa
Deposited On:04 Mar 2007 14:53
Last Modified:28 Aug 2017 04:08

Repository Staff Only: item control page