Information Technology Reference
In-Depth Information
4.3 Subjective Quality Results
To determine whether the performance gains shown in the objective results are sig-
nificant, the quality of the proposed method has been evaluated subjectively for the
'Interview' sequence. This sequence is chosen for subjective tests because the gains
in objective quality are the lowest of the four tested sequences. The test used the
single stimulus method described in ITU-R BT.1438 recommendation, which spec-
ify the quality evaluation procedure for stereoscopic video applications. A 42” Phil-
ips multi-view auto-stereoscopic display was used in the experiment to display the
stereoscopic material to 16 subjects. Subjects rated the test sequences for the over-
all perceived image quality. The Mean Opinion Scores (MOS) have been calculated
for each test sequence after averaging the opinion scores across all subjects.
Table 3 shows the subjective test result for the proposed prioritization scheme
compared to the non-prioritized case. The MOS difference is greater than the 95%
Confidence Interval (CI). This shows that a perceptible difference in quality can
be observed with the proposed technique, even for sequences that exhibit a small
gain in objective quality.
Table 3
Subjective Comparison for the Interview Sequence for Test Case A, D
Un Prioritized ± CI
Prioritized ± CI
Err Free
3.59 ± 0.392
3.59 ± 0.392
Case A
2.67 ± 0.324
3.54 ± 0.352
Case D
2.06 ± 0.308
2.83 ± 0.376
5 Conclusions
A novel object based prioritization scheme for optimized transmission of 3D color
plus depth video content over wireless networks is proposed in this work. A 3D
home entertainment scenario is considered and an efficient channel resource allo-
cation scheme is implemented for improved visual quality. Video content is seg-
mented into video objects, based on the depth map, and the expected distortion is
then calculated at the video packet level using the conventional 2D distortion pa-
rameters plus the effective contribution of depth information to distortion for each
object. Data prioritization is achieved by rearranging the video packets into
different bit streams, according to estimates of their distortion. These bit streams
are then transmitted over prioritized communication links with different QoS fa-
cilities supported by the WLAN communication standard, IEEE 802.11e. The ex-
periments carried out to evaluate the proposed scheme have demonstrated
