Conversion of 3D Video Data

Background

3D video is gaining momentum and soon mobile phones and home TV will be 3D ready. Various display technologies are competing for market domination, ranging from stereoscopic displays with glasses (anaglyphic, shutter and polarized) to auto-stereoscopic. The latter case does not use glasses. Instead two different images are displayed and the viewer needs to position herself in a specific location in front of the device so that different images arrive on her left and right eye respectively. An extension to the auto-stereoscopic display is the so-called N-view auto-stereoscopic displays where multiple viewers simultaneously can experience the stereo effect without using glasses.

An open issue for 3D video is the lack of content and the increased cost for content generation. Existing and new techniques should be investigated towards 2D-to-3D conversion and stereo to multi-view conversion. The areas of interest range from depth estimation (depth and structure from motion, stereo or multi-views), depth improvements, view rendering, etc.

Task

The goal of this thesis is to study, implement, develop and evaluate different 2D to stereo to multi-view algorithms. The thesis will contain a theoretical study, design and implementation and a technical report.

Competence

We are looking for two bright M.Sc. students with interests in video processing and programming.
The candidates should have general knowledge on image/signal processing and computer vision. Knowledge in machine learning would be an advantage. Good programming skills in C/C++ and Matlab are required.

Ericsson Research – Visual Technology

The thesis work will be done at the Visual Technology unit at Ericsson Research in Kista outside Stockholm. The unit is responsible for research related to video coding, image coding, graphics and image processing for multimedia communication and distribution.

Applications should contain the following documents: a short personal letter, recent grades (undergraduate and master studies) and a CV. Candidates are invited to send their applications as soon as possible.

Apply online before 2010-08-15 by following the link below. Applications by e-mail are not accepted.

Contact persons

Apostolos Georgakis, Ph.D.
Research Engineer

Thomas Rusert, Ph.D.
Senior Research Engineer

Per Fröjdh. Ph.D.
Manager, Visual Technology