A non-intrusive video tracking method to measure movement of a moored vessel

Kieviet, Johan (2015-03)

Thesis (MSc)--Stellenbosch University, 2015.

Thesis

ENGLISH ABSTRACT: There are several ports around the world currently experiencing problems with moored vessel motions. Extreme vessel motions are mainly caused by long waves, which can become trapped inside a harbour basin. The extreme motions can cause downtime in port operations and in some instances cause mooring lines to break. Methods and procedures currently available to measure motions of moored vessels require vessel specific information as input. The implementation of these methods is seen as impractical to implement on every vessel visiting the port and require the physical measurement of some points on the vessel and/or the placement of some kind of measurement device on the vessel. A new Six Degree of Freedom (6DOF) motion measurement system for a moored vessel is presented in this document. The system analyses a video image sequence from one camera. The method estimates the 3D rigid motion for an object of known size by using a Pose from Orthography and Scaling with ITerations (POSIT) algorithm. The object for which the motion is estimated is located on the deck of the vessel and within the camera field of view. Geometric rigid body calculations allow for the calculation of camera perspective rotations and translation of an object on the vessel. Further geometric calculations allow for converting camera perspective motions to the 6DOF object motions. The primary objective of this study was to validate and verify the motions obtained from two proof-of-concept tracking systems. For evaluation purposes, the validation was done by using a small scale physical model set-up in a hydraulics laboratory and using a known method as reference. The Keoship system from the Council for Scientific and Industrial Research (CSIR) is currently one of the most accurate small scale vessel motion measurement systems and was used as reference. The first method tested was the tracking of a 2D LED rectangle mounted on the vessel. This method tracked a 2D object and was primarily used as a stepping stone to measure movement of a 3D object. The second method tracked a 3D object on the vessel. Each tracking method was tested for four different wave conditions with each condition additionally repeated twice as repeatability tests, resulting in a total of 12 tests for each tracking method. When comparing the 2D LED tracking and 3D Object tracking data to data measured with the Keoship system, results show that in general, the 3D Object tracking data compared better to the Keoship data. Tests under controlled conditions enabled a direct estimation of the absolute accuracy of the two developed methods. The verification and accuracy test results, indicated that the 2D LED tracking system should not be pursued further. The results also indicated that for prototype motions exceeding 0.6 m (i.e. storm events) the 3D Object tracking system would have an accuracy close to the maximum allowable accuracy criterion of 0.1 m. This makes the system viable at its current proof-of-concept stage for further development which would enable rapid deployment during a storm event in a prototype situation.

AFRIKAANSE OPSOMMING: Daar is verskeie hawens regoor die wêreld wat tans bewegings probleme op gemeerde skepe ervaar. Hierdie buitensporige bewegings word veroorsaak deur lang periode golwe wat binne die hawe bekkens vasgekeer word. Dit kan daartoe lei dat hawe bedrywighede tot stilstand kom en in ernstige gevalle ook veroorsaak dat meringslyne breek. Huidige metodes vir die meet van skeepsbewegings op vasgemeerde skepe, vereis skeep spesifieke inligting as inset. Die toepassing van hierdie metodes op elke skip wat die hawe besoek, word as onprakties beskou, aangesien dit die fisiese meting van sekere punte op die skip behels. In sekere gevalle is dit selfs nodig om meet toestelle op die skip te plaas. In hierdie dokument word ‘n nuwe metode aangebied om die ses grade van vryheid bewegings vir ‘n vasgemeerde skip te meet. Hierdie stelsel analiseer ‘n video beeld reeks van een kamera. Die metode bereken die 3D rigiede beweging van ‘n voorwerp, waarvan die grootte bekend is. ’n ‘Pose from Orthography and scaling with Iterations’ (POSIT) algoritme word hiervoor gebruik. Die voorwerp waarvoor beweging gemeet word is op die dek van die skip en in kamera sig. Rigiede geometriese voorwerp berekeninge word gebruik om die rotasie en translasie vanuit ‘n kamera perspektief te bereken. Verdere geometriese berekeninge maak dit moontlik om die bewegings vanuit die kamera perspektief te omskep in die ses grade van vryheid bewegings van die voorwerp. Die hoof doelwit van hierdie ondersoek was om die gemete bewegings van twee beweging stelsels te valideer en te verifieer. Die validasie en verifiëring was in ‘n hidrolise laboratorium met ‘n klein skaal model opstelling getoets. ‘n Meet metode van skeepsbeweging op klein skaal wat reeds bekend is, is gebruik as ‘n verwysingsraamwerk waarteen die metings vergelyk kan word. Die Keoship stelsel van die Wetenskaplike Nywerheids Navorsings Raad (WNNR) is tans een van die mees akkurate klein skaal skeepsbeweging meet stelsels, en was as verwysing gebruik. Die eerste bewegings metode is getoets op ‘n 2D reghoek vervaaridig uit ligstralede diodes. Hierdie metode het die 2D voorwerp gevolg en is hoofsaaklik gebruik as ‘n boublok om die beweging van ‘n 3D voorwerp te volg. Die tweede metode het die beweging van ‘n 3D voorwerp op ‘n skip gevolg. Vir elke meet metode was daar vier verskillende golf toestande. Elke golf toestand was ook ‘n verdere twee keer herhaal vir herhaalbaarheids doeleindes. Saam met die herhaalbaarheids toetse was daar in totaal, 12 toetse vir elkeen van die twee metodes gedoen. Met die Keoship metode as verwysing, bewys hierdie toetse dat die 3D metode beter resultate lewer as die 2D metode. Toetse onder beheerde toestande, het dit moontlik gemaak om die absolute akkuraatheid van albei sisteme wat ontwikkel was, te evalueer. Verifikasie en akuraatheids toetse het aangedui dat verdere ontwikkeling van die 2D metode gestuit moet word. Die resultate het ook aangedui dat die 3D metode ‘n akuraatheid baie na aan die maatstaf van 0.1 m sal hê wanneer prototipe bewegings 0.6 m oorskrei (b.v. gedurende ‘n storm). Dit sal die oplossing lewensvatbaar maak by die huidige bewys van konsep fase vir die verdere ontwikkeling wat vinnige ontplooiing gedurende ‘n storm sal moontlik maak.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/97074
This item appears in the following collections: