Image Processing Reference
In-Depth Information
10.3.2.5 Viewers
Viewer is another important influence factor on the experimental results as the
gender, the age, or the viewing experience of the viewers would affect the exper-
imental results. Viewer
s information should be well documented and these factors
should be balanced. For example, the gender and the age should be approximately
uniformly distributed.
In this study, 42 viewers participated in Experiment 1. 21 are male, 21 are
female. They are all non-experts in the domain of subjective experiments, image
processing, or 3D. Their age ranged from 19 to 48 with an average age of 26.8. Ten
experts in 3D perception, coding, quality assessment, and subjective experiments
participated in the Experiment 2-a. Eight experts are male, two are female. Their
ages ranged from 24 to 43 with an average age of 27. As the number of viewers is
too small, to generate a reliable result, 5 of them conducted the test twice but on a
different day. Thus, for each pair, there are 15 observations in total. 45 naive
viewers participated in the Experiment 2-b. Twenty-one are male, 24 are female.
They are all non-expert in subjective experiment, image processing, or 3D related
field. Their ages ranged from 18 to 44 with an average age of 24.
All of the viewers have either normal or corrected-to-normal visual acuity. The
visual acuity test was conducted with a Snellen Chart for both far and near vision.
The Randot Stereo Test was applied for stereo vision acuity check, and Ishihara
plates were used for color vision test. All of the viewers passed the pre-experiment
vision check.
'
10.3.2.6 Assessment Method
The paired comparison method was used in this test. As there are 36 stimuli in
Experiment 1, to reduce the number of comparisons, the ASD method was used
[
42
]. In Experiment 2, as only 15 stimuli were considered, the FPC method was
used. In Experiment 1, according to ASD method, 36 stimuli lead to a total of
180 pairs for each viewer. In Experiment 2-a, only 10 experts conducted the
experiment. In order to produce more reliable results, each viewer conducted the
test using the FPC (Full Pair Comparison) method and both presentation orders for
each pair were considered, i.e., each stimulus pair will be shown twice but with
different order, the stimulus which is shown firstly in one trial will be shown
secondly in another trial. Thus, there were in total 210 pairs for each viewer in
Experiment 2-a. In Experiment 2-b, a total of 105 pairs were presented to each
viewer. The presentation order of the stimulus pair was different for odd numbered
and even numbered viewers. For example, viewers with even numbers will watch
stimulus
A
first, then stimulus
B
. For odd numbered viewers, this order is inverted.
This is used to balance the presentation order. The presentation order of each
stimulus pair in all experiments was randomly permuted for each viewer.
