Underwater reconstruction and localisation for ship hull inspection
Time & Place
Fri, 24 Nov 2017 13:00:00 NZDT in Erskine 315
All are welcome
The autonomous visual inspection of ship hulls or the monitoring of aquacultures with an underwater vehicle is an active field of research. Compared to manual inspections by divers, autonomous systems enable safe, cost efficient, and more comprehensive inspections of underwater structures. Additionally, the collected data enables the automatic 3D reconstruction and assessment of the inspected areas for the detection of unwanted marine organisms and structural damages and monitoring mussel farms.
Accurate high-resolution (0.1mm) colour 3D reconstruction of underwater surfaces is needed to reliably recognise invasive marine species. Due to turbidity, cameras need to be close to the surface (with 1m) and so very wide angle lenses are needed to cover a large enough hull area for each scan. But at such large angles and colour camera resolutions, red light can refract up to 13 pixels away from the blue pixels depending on pressure, salinity and temperature.
In this presentation, we will give an overview of our ongoing research on accurate and fast underwater 3D reconstruction and self-localisation to solve the above and other issues. The stereo camera used in our research is embedded in an underwater housing with a flat window (port). Due to the refraction of light, 3D reconstruction and localisation cannot be fully handled with classical in-air lens or perspective camera models. The presentation will give an insight into the effects caused by refraction in multi-media wide-angle flat refractive geometry, and discuss our achievements solving fast refractive projection, accurate 3D reconstruction, and accurate real-time self-localisation.
"Note that this seminar is sponsored by the IEEE New Zealand Computer Society"
Robert Schattschneider received his MSc degree in computer engineering, specialising in robotics and embedded systems, from the Technical University of Berlin, Germany, in 2006, and his PhD degree in computer science from the University of Canterbury, New Zealand, in 2015, where he was a postdoctoral researcher in the Department of Computer Science and Software Engineering. His current research involves the 3D reconstruction and localisation with a wide-angle flat port underwater stereo camera.