Image Processing Reference
In-Depth Information
Fig. 3.7 Example of LDV
representation, such as additional texture and depth maps for pixels that are
occluded in the transmitted view. This extra data, shown in the bottom pictures of
Fig. 3.7 , provide the necessary information to fill in the disoccluded regions of the
newly generated view.
This representation can be an alternative to MVD that contains a lot of redundant
information, as foreground and occluded areas will be visible from multiview
points. Thus, the LDV image representation may transmit only one view (with
associated depth map) and additional residual data from the occluded areas, thus
reducing the required bitrate [ 18 ]. This scheme may be extended to include two sets
of LVD, which is referred to as depth-enhanced stereo (DES) [ 19 ]. This represen-
tation format is more suited for stereoscopic glasses-based consumer displays,
increases the occlusion information, and allows interpolation or extrapolation of
the views.
3.3
3D Video Coding
Despite the potential of the stereoscopic and MVV formats, for a variety of
applications, all of these formats require a much larger amount of data than the
traditional 2D representation of a scene. The escalation of the requirements asso-
ciated with storage and transmission of these sequences has motivated the devel-
opment of many different algorithms to encode 3D video. In this section, we present
an overview of the most important coding algorithms for 3D video sequences, using
the representation formats described in Section 2.
Search WWH ::




Custom Search