A computer-vision based automated underwater ranging system

Meek, Lindsay (1996) A computer-vision based automated underwater ranging system. Masters by Research thesis, Murdoch University.

Abstract

This thesis describes a ranging system that uses binocular stereoscopic images to estimate the position of objects in an underwater environment. These positions can be judged by an operator using interactive computer graphics, or automatically using a target tracking system. The target tracking system follows the movement of objects using image processing techniques. This system produces position estimates for the objects every 350 ms with an error of less than ±0.1 metres. The typical operating range of the tracking system is 1-2 metres.

The thesis may be divided into four main sections. The first section describes the development of the target tracking system. This system was realised using a correlation-based feature-matching process. The software for the tracking system was implemented using a data-flow parallel architecture. A novel software management CASE tool was created to simplify the development of this software. Another novel aspect of the tracking system was the development of a process to provide station keeping to the underwater vehicle. A series of coordinate system mappings were required to translate the observed movement of objects to vehicle movements.

The second section of the thesis describes the hardware used to implement the ranging system. This hardware consisted mainly of a 3DTV video system, developed at Curtin University. This system was extended to incorporate an overlayed 3D computer image, using off-the-shelf imaging hardware and a novel overlay technique. This technique required the development of a calibration method which corrected the effects of the timing mismatch between the two video signals.

The third section of the thesis describes the implementation of an interactive ranging system for remote teleoperation. This system uses a mouse-controlled 3D pointer, as described by Drascic. The technique has been extended to incorporate stereo matching to improve the accuracy of the depth estimates.

The fourth section of the thesis describes the stereo triangulation process and its calibration for a given camera configuration. The system uses a homogeneous coordinate mapping, as described by Sid-Ahmed. This approach has been extended to include a method of gathering calibration data in an automated and accurate fashion. It has also been extended to provide the extrinsic camera parameters for a given camera model. These parameters can be used to verify calibration results or allow the simpler camera model to be used. A series of equations have been derived to provide these parameters.

The ranging system was subjected to various tests to determine its performance both in the laboratory and in the field. The results of these tests are presented and discussed, followed by recommendations for improvements to the system.

Item Type:	Thesis (Masters by Research)
Murdoch Affiliation(s):	School of Physical Sciences, Engineering and Technology
Notes:	Note to the author: If you would like to make your thesis openly available on Murdoch University Library's Research Repository, please contact: repository@murdoch.edu.au. Thank you.
Supervisor(s):	Koch, Rolf
URI:	http://researchrepository.murdoch.edu.au/id/eprint/51664

Item Control Page