A group of postgraduate students and researchers from the Faculties of Science and Engineering are on their way to Europe for two months as part of an international exchange scheme.
Prof. Ben Herbst, head of Applied Mathematics in the Department of Mathematical Sciences at SU, says during the visit they will meet with their counterparts to work on a project called QOSTREAM. Funded by the European Union, it involves researchers from eight universities in Europe, the USA, the UK, Brazil and Japan.
The aim of the project is to combine expertise from different fields to solve problems associated with high quality multimedia streaming applications across next-generation wireless networks. The project was initiated by Prof. Dibravko Culibrk and Prof. Vladimir Crnojevic from the Faculty of Technical Science at the University of Novi Sad, Serbia.
The group from Stellenbosch University will bring their expertise in machine learning, computer vision and signal processing to the table. The students will be accompanied by Prof. Herbst and Dr Riaan Wolhuter, senior researcher in the signal processing and machine learning research group in the Faculty of Engineering.
"We are providing the more theoretical foundation on which many of these new technologies are based," Prof. Herbst explains.
In other words, your average handheld mobile device requires some cool algorithms to decode the information it receives via wireless networks. By exploring different approaches and techniques to optimise this process, the researchers hope to maximise the quality of images and videos received via mobile devices.
"We are nowhere near the type of technologies showed in popular television series like CSI. They are typically given a blurry photograph and then asked to extract additional, not visible, information from it. With some knowledge about the degradation, one could extract some information from the photograph. But what is shown in CSI cannot be done in reality, not now, not ever.
"Reality is much more complicated. We are currently playing around with novel approaches such as layered media coding, deep learning and neural networks," he explains.
"Applied mathematicians like working with problems for which there are no answers yet," he adds.
The postgraduate students are all working on different aspects of this problem. They are Andrea Fourie and Marco Grond (applied mathematics), Jacques Marais (MIH Media Lab), S.P. le Roux, M.N. Struwig, W. Pretorius and J.P. Meijers (signal processing and machine learning research group).
Other SU researchers involved with the project are Prof. Johan du Preez, Dr Pieter Holtzhausen and Mr Willem Smit.
