| Person in charge of the project: |
|
Interactivity is an important key feature of new and emerging audio-visual media where the user has the opportunity to be active in some way instead of just being a passive consumer. One kind of interactivity is the ability to look around within an audio-visual scene by freely choosing viewpoint and viewing direction. It allows free navigation within dynamic real-world scenes by choosing arbitrary virtual viewpoints and viewing directions. The 3D acquisition typically relies on a multi-camera setup and an event is captured by a given number of real cameras. The images of such a multi-view capturing unit are then converted into a suitable 3D data representation format, which supports the functionality to interpolate intermediate views at arbitrary camera positions. Free viewpoint television and 3D video conferencing - where the user can freely interact with their peers in a 3D world as if they were there - are just small examples of what can be achieved with Multi-View Video technology. The technological challenges that are addressed in this PhD study are:
# MultiView Video Compression: Achieving high video camera compression ratios by exploiting inter-camera redundancies in a distributive manner
# MultiView Video Quality of Experience: Achieving high wireless transmission error immunity by advanced multi-resolution video coding and scheduling techniques
# Interlinking MultiView Video Compression and Reconstruction: Bridging the gap between the "motion estimation" techniques, typically used in video compression, and "feature extraction/correlation" techniques used for camera calibration in multi-view video, hence moving/distributing the free viewpoint reconstruction algorithm to/over the processing power available at the cameras (possibly a massive amount of tiny "intelligent" cameras distributed over a huge area, e.g. city).
# MultiView Video Image Processing: Providing low-cost implementation solutions to low-level image processing algorithms (e.g. occlusion detection, boundary matting, #), which are essential for multi-view video reconstruction
|
MAINALI PRADIP
Faculty of Engineering
Doctoral Programme in Engineering
|
| |
Project number:
3E090291
Duration of the project:
27.10.2008 - 27.10.2012
Funded research
Nederlands
|
