Quality Assessment - 3DTV: Processing and Transmission of 3D Video Signals - page 167

6.3.1.1 Subjective Test Design
The experiment has a within subjects design, with several colour-plus-
depth-based 3D video sequences and different encoding combinations as
independent variables and 3D quality attribute(s) as dependent variable(s).

6.3.1.2 Selection and Screening of Observers
Observers are usually either classified as being expert or non-expert viewers.
Expert viewers will usually give more consistent results, but their opinions
may not be representative of those of the general public. They are often very
good at recognizing artefacts within video compared to non-expert viewers.
For this reason, when expert viewers are used, a lower number of observer
are used (e.g. 5 - 10 viewers). If non-expert viewers are employed, then it is
necessary to carry out a much larger number of tests to ensure the reliability
of results, and to be able to exclude outliers.
For the tests in this chapter, 28 non-expert observers volunteered to
participate in all experiments. The observers are mostly research students
and staff with a multimedia signal processing background. Their ages range
from 20 to 40. All participants had a good visual acuity ( > 0.7, as tested with

a Snellen eye chart), good stereo vision ( 60 seconds of arc, as tested with

the TNO stereo test) and good colour vision (as tested with the Ishihare test).
Each assessor is well informed on the test process and the test materials
(possible quality defects).
Eye dominance refers to the tendency of the brain to prefer visual input
from one eye over the other. The 'Finger-Point' method is often used to
determine eye dominance. In this method, observers are pointed natu-
rally at an object with both eyes open and the face square to the object.
The eyes are closed alternately. When the dominant eye is closed, the
finger appears to jump away from the original location. Eye dominance
is particularly important when evaluating techniques, such as asymmetric
compression, where the left and right eyes are compressed with different
qualities.

<

6.3.1.3 Displays
The 42'' Philips WOWvX multi-view auto-stereoscopic display (display
resolution-1920

1080, aspect ratio-16:9 and the peak luminance of the
display is 200 cd/ m 2 ) is used in all experiments to display the 3D content.
The advantage of this display, besides 3D viewing without glasses, is the
support of motion parallax enabling the viewer to look around objects
by moving their head. The viewing distance for the observers is set at
3m, which is optimum for the display optics. The 3D display is calibrated
using a GretagMacbeth Eye-One Display 2 calibration device. The measured
environmental illumination is 190 lux, which is closer to the recommended
value in [11] for home environments (i.e. 200 lux). The background luminance

×