Project R-4583

Title

ASPRO+: AutoStereoscopic multi-view video PROcessing and PROjection

Abstract

Stereoscopic 3D became important in a growing number of fields. 3D movies are commonplace nowadays. However, for many other applications (like 3D advertisements and events, 3D entertainment, 3D surveillance, 3D gaming, 3D communication, ...), stereoscopic 3D still remains an immature technology with a number of limitations. Many experts believe that the future of 3D lies in a much more sophisticated 3D technology, with more 3D viewpoint freedom. Nevertheless, quite some technical hurdles needs to be solved before this can be brought to the market: A first hurdle is related to the 3D visualization. Almost all stereoscopic 3D displays have either the obligation to wear specialized glasses, or require the viewer to be located at a very precise sweet spot. So-called autostereoscopic 3D displays and projection systems are seen as the solution for this visualization problem. A second hurdle is related to the 3D acquisition. Stereoscopic 3D is typically captured with two cameras. Stereoscopic camera rigs are expensive and not portable enough. Further difficulties are faced in synchronously controlling focus and zoom of the camera pair. These issues become even more problematic when going for an autostereoscopic 3D capturing, which requires a large array of cameras. A third hurdle is related to the 3D processing of the content. On the one hand there is a huge processing required in order to generate the right content for each projector, based on the enormous amount of data captured by the camera array. On the other hand, a lot of representation formats exist for 3D nowadays, e.g. MVC, 2D+Z, MVD, ... However, no optimal 3D representation format is available that supports all aspects of the acquisition-reproduction pipeline, allows for seamless free viewpoint views, and is at the same time rendering (and possibly editing) friendly. A last hurdle is related to detecting and tackling 3D quality issues, especially regarding spatio-temporal quality, depth perception and naturalness. ASPRO+ will perform research in this area, taking into account contextual factors such as screen size and ambient lighting. The goal of ASPRO+ is to explore the technological challenges and provide solutions for the current limitations throughout the entire autostereoscopic 3D video processing chain, i.e. from acquisition to display. This will result in a functional proof of concept for an autostereoscopic multi-view 3D imaging workflow.

Period of project

01 January 2013 - 31 December 2014