Digital Signal Processing Reference
In-Depth Information
(a)
(b)
Figure 1.8
Example colour-plus-depth video sequence commonly used by
video coding experts for testing video compression algorithms, called Inter-
view, produced in the course of ATTEST Project (a) Colour sequence. (b) Depth
sequence (Reproduced with permission of the ATTEST Consortium)
Depth-Image-Based (DIBR) can be used to synthesize two views for the
left and right eyes using colour image sequences, and the corresponding
per-pixel depth map [37, 38]. This process requires two main steps: re-
projection of original image point into 3-D space using depth information,
and projection of the 3-D space points into the image planes of the left and
right views.
The advantages of using colour-plus-depth map representation of stereo-
scopic video compared to video generated with a stereo camera pair can be
summarized as follows:
•
3D rendering can be optimized for different stereoscopic displays and
viewing scenarios to yield a disparity that is comfortable to the eye.
•
Head-Motion Parallax (HMP) may be supported, which provides an
additional 3D depth clue. This format also partially overcomes the viewing
angle limitation of traditional stereoscopic camera set-ups.
•
Most depth information does not have high frequency components. Thus,
the depth sequence can be efficiently compressed with existing compres-
sion standards [39], and will require only limited space and bandwidth
compared to that required by the colour image sequence.
•
Photo-metrical asymmetries (e.g. in terms of brightness, contrast or colour)
between the left and the right eyes, will be eliminated. Thus, the associated
eyestrain problems will be avoided.
•
Depth can be used in 3D post production (e.g. fine tuning of depth to
eliminate stereoscopic artefacts that may occur during filming).
However, there are a number of drawbacks associated with this represen-
tation. The disadvantages and possible solutions are as follows:
Search WWH ::
Custom Search