On Seeing Transparent Surfaces

1
On Seeing Transparent Surfaces in Stereoscopic
Displays
Inna Tsirlin-Zaharescu, Robert S. Allison and
Laurie M. Wilcox
2
Motivation

• Stereo-transparency presents a difficult problem
  both theoretically and computationally

• Small number of psychophysical studies addressed
  stereo-transparency (Akerstrom Todd 1988,
  Gepshtein Cooperman 1998, Wallace Mamassian
  2004)

• Goals

• Investigate the number of overlaid planes
  limiting the perception of stereo-transparency

• Study the combined effects of density, disparity
  and number of planes on the percept of
  stereo-transparency.

3
Methods

• Stimuli were 12.6o square RDSs depicting n
  overlaid planes
• of dots with inter-plane depth separation d.

• In Glass planes dots were arranged in a
  horizontal-shift Glass pattern.

• Task the observers were asked to report the
  number of planes and the number of Glass planes
  they perceived.

• Stimulus parameters
• Density 2, 6.3, 10, 15.7 dots/deg2
• Inter-plane separation 1.9, 3.8, 5.7, 7.6
• Number of planes 2, 3, 4, 5, 6

4
DEMO
5
(No Transcript)
6
Discussion
Possible explanation
Detrimental effects on stereo-transparency
Disparity pooling
Small disparities
Size-disparity correlation Adapted from
McKee Verghese 2002
Increase in disparity beyond an optimal value
Increase in density

• Increase in the number of planes

7
Density-disparity interaction
Akerstrom and Todd (1988) Wallace and Mamassian (2004)
Effect of increase in disparity Performance decreased No effect
Stimuli parameters 28.7 dots per square degree 7.08 dot size 112 max. disparity 8.9 dots per square degree 4.5 dot size 60 max. disparity
8
Conclusions

• We are able to segregate up to 3 planes under all
  conditions and up to 5 planes under optimal
  conditions.

• Increasing the number of planes, density and
  disparity beyond an optimal value degrades
  performance

• The interaction of these parameters plays an
  important role in stereo-transparency and cannot
  be ignored in future research

• Stereoscopic modeling based on a combination of
  position-shift and phase-shift energy detectors
  cannot account for the experimental results.

9
Thank you!
Questions?
10
Motivation

• Stereo-transparency presents a difficult problem
  both theoretically and computationally

• Small number of psychophysical studies addressed
  stereo-transparency

• Used RDS stimuli with only 2 overlaid planes

• Varied RDS parameters in isolation

11
Types of stereo-transparency