Quality Assessment - 3DTV: Processing and Transmission of 3D Video Signals - page 188

Two view
stereoscopic

Multi-view Light-field

Figure 6.19 Typical Field of View for 3D display technologies

a single view of the display using a mid-grey background while measuring
the brightness of the screen at its neighbouring views. This and the view
uniformity can also be measured using Eldim EZLite3D.

6.4.6.5 Comfort
Some users may not be happy to wear filter glasses. Polarization-based
techniques also require glasses. Depending on type of polarization, this
technique may also prevent the user from tilting his/her head. Although
shutter glasses-based 3D displays provide a full resolution 3D image, which is
considered important, they usually cause headaches for most users, especially
if the refresh rate is not sufficiently high. This severely affects the users'
comfort, even when the image quality and depth perception are sufficiently
good. The auto-stereoscopic technique reproduces colour images on both
eyes without any support from wearable visual aids. However, correct
3D vision in most auto-stereoscopic displays is limited to special angles,
which are identified as 'sweet spots'. Even if the viewer can see the correct
image by holding his/her head strictly in the sweet spot, he/she may
feel uncomfortable after a while. Therefore, such restrictions may prevent
the viewer from enjoying 3D visual contents. However, modelling this is
extremely difficult without proper subjective tests.

6.5 Conclusion

3D video in the format of monocular video (colour texture video) augmented
by the greyscale depth map is a promising representation technique, which
provides high quality 3D video with a smaller bandwidth. This chapter has
addressed the possibility of modelling important attributes of the above
format of 3D video, which could accelerate the further development of
broadcast technology and the introduction of new 3D video services in time.
First, subjective results are analysed to investigate the effects of monocu-
lar colour texture video and the depth map on image and depth quality
attributes as experienced by the viewers. Second, the possibilities of mod-
elling these attributes are investigated. To predict the image quality of 3D
video, three widely used 2D quality models have been considered with four
different combination approaches incorporating different aspects of visual