With its Best Paper Award 2014 for “The effects of multiview depth video compression on multiview rendering”, the European Association for Signal Processing (EURASIP) honors the work done by Fraunhofer HHI on the special characteristics of depth maps in the context of 3D video coding, with a particular focus on distortions resulting from coding artifacts and their impact on rendered intermediate views.
The paper investigates standard video coding techniques (H.264/MPEG-4 AVC) and their multiview extension (MVC), in whose development Fraunhofer HHI has played a leading role. Furthermore, it presents a new coding technique adapted to the special characteristics of depth maps, which is based on geometric modeling functions, so-called “platelets”.
In contrast to classical video coding, distortions or errors in depth maps degrade the visual quality indirectly through the distortions they cause in rendered video views. Consequently, the paper investigates the effect of depth map distortions on the resulting visual or subjective quality, with particular focus on suitable quantitative analysis methods.
The results show that platelet-based coding of depth maps leads to a considerably higher subjective quality of rendered intermediate views compared to H.264-based depth coding. Furthermore, a comparison of the investigated quality analysis methods shows that a reliable quantitative assessment of the resulting subjective quality is only possible by measuring the distortion in rendered video views.
The paper was developed in collaboration with the group of Professor P.H.N. de With at Eindhoven University of Technology. The approaches and methods described in the paper represent significant preliminary work for the 3D video codec subsequently developed by Fraunhofer HHI, which was a winning submission in MPEG’s Call for Proposals. Based on this submission an extension of the HEVC standard for 3D video coding is currently being developed and is expected to be finalized in 2015.
Signal Processing: Image Communication
P. Merkle, Y. Morvan, A. Smolic, D. Farin, K. Müller, P.H.N. de With, T. Wiegand, “The effects of multiview depth video compression on multiview rendering,” Signal Processing: Image Communication, Vol. 24, Nos. 1–2, January 2009, pp.73-88.