Stereo vision, 3D video capture and scene representations - 3D Video Processing and Transmission - page 37

However,theexistingdrawbacksofthisrepresentation,hasledtoseveralresearchfindings
which can be utilized to mitigate the effects of the monoscopic video plus depth map re-
presentations.Thedisadvantages of using this representation and the solutions come across
are listed as follows.
• The quality of the rendered stereoscopic views depends on the accuracy
of the per-pixel depth values of the original imagery. Therefore, the effects of
compression and transmission of depth maps (e.g. introduced artefacts) on the
perceived quality need to be thoroughly investigated.
• The visible objects for the rendered virtual left and right views may occlude
from the original view. This phenomenon is also known as exposure and
disocclusion in Computer Graphics (CG) [21]. This effect can be minimized us-
ing Layered Depth Images (LDI) where more than one pair of colour plus depth
sequences is transmitted depending on the requirements of the expected quality
[26]. However, this approach demands more storage and bandwidth to be used
in communication applications. In addition, different hole-filling algorithms (e.g.
linearinterpolationofforegroundandbackgroundcolour,backgroundcolourex-
trapolation, mirroring of background colour information) can be utilized
to recover the occluded areas of the original image sequence [24]. Moreover, the
pre- processing/smoothing of depth maps (e.g. use of a Gaussian filter) will avoid
this occlusion problem. However, this approach will lead to some geographical
distortions of the rendered 3Dvideoscene.
• Certain atmospheric effects (e.g. fog, smoke) and semi-transparent objects
aredifficultto handlewiththisapproachatthemoment.
• The processing overload (e.g. memory, processing power, storage
requirements) at the receiver side is high compared to the reconstructing 2D
video stream.
The monoscopic video plus depth map representation is widely utilized in research and
standardization activitiesduetoitssimplicityandadaptability [27] [28] [29]. TheATTEST
(AdvancedThree-Dimensional Television System Technologies) project consortium is
workingon3D-TVbroadcasttechnologiesusing colour-depth sequences as the main source
of 3D video [27]. Recently, ISO/IEC 23002-3 (MPEG-C part 3) finalized the
standardization of video plus depth image representations/solutions in order to provide:
interoperability of the content, flexibility regarding transport and compression techniques,
displayindependenceandeaseofintegration [28]. Moreover,JVThasidentifiedmulti-view