Cebit 2017: Jump Into Your Favorite Movie Scenes With Multi-View Video Streaming

Possible applications range from consumer electronics, computer games and virtual worlds to realistic video conferences and autonomous driving. Until now, there was no overall functioning system for this technology. Now computer scientists from Saarland University will be presenting their solution at this year’s Cebit computer fair.

Thanks to advancements in camera technology, a single scene can be captured from different perspectives with several cameras at once. This opens up entirely new fields of application, especially for home electronics: “Say you are watching a detective story, and you can hear the suspect talking in the room next door, off-screen. Then you could use this new technology to simply switch perspectives and look around the corner,” says Tobias Lange, PhD student at the Saarland University.

Together with Thorsten Herfet, Chair of the Telecommunications Lab at Saarland University, Tobias Lange is researching multi-view video streaming at the Intel Visual Computing Institute. The individual components of this technology already work by themselves, but so far there has been no complete integrated system. This is what the two Saarland computer scientists and their colleagues are trying to change. The potential applications for multi-view video streaming are abundant.

Multi-view video streams could not only revolutionize the entertainment industry, but also the workplace, with true-to-life video conferencing, for instance. “This technology might even be useful if several autonomous cars are driving in convoy. Or how else would the car at the rear be able to access images from the car in front?,” asks Lange.

Presently there are still some challenges that have to be addressed in order to fully implement this vision. The greatest difficulty: Network technology has not developed as rapidly as recording technology has. Not only do the videos need to be recorded with several cameras, they also have to be encoded into a set of data packets and stored in a transmission buffer on the video server.

From there, they will find their way over the Internet to viewers’ computers. There the data needs to be unpacked in time and played back in such a way that viewers can switch their perspective without encountering any image errors. “The data rate is ludicrous. Even now we need such a high bandwidth that most contemporary Internet connections would be overloaded,” Lange explains. The calculations themselves are also highly complex.

The Saarbrücken researchers are tackling this challenge by improving the entire production process step by step. For streaming, they use special computational methods to minimize the delay until it is so small as to be nearly unobservable. And for encoding and decoding the data in an acceptable time span, they rely on a distributed approach. Each camera has a mini-computer of its own attached to it.

The researchers have also developed a new method for computing the images that can be viewed from different perspectives almost in real-time. With these incremental improvements, they have created a comprehensive working system. “Only a few solutions exist,” says Lange. Hence, he is unsure whether that is sufficient to make the technology marketable.

“Maybe our research partner Intel will take over. They will be looking at our results in their entirety,” says Lange. The researchers are now presenting their completed system at the Cebit computer fair from March 20 to 24 in Hannover (Hall 6, Stand E28).

Additional Information:
http://www.intel-vci.uni-saarland.de/de/projekte/efficient-multi-view-video-stre…

Press photos are available here: www.uni-saarland.de/pressefotos

Further Inquiries:
Thorsten Herfet
Telecommunication Lab
Saarland Informatics Campus C6 3
Saarland University
Phone: +49 (0)681 302- 70852
E-Mail: herfet@cs.uni-saarland.de

Editor:
Gordon Bolduan
Competence Center Computer Science Saarland
Saarland Informatics Campus E1 7
Phone: +49 681 302-70741
E-Mail: bolduan@mmci.uni-saarland.de