3D Surface Configuration Modulated 2D Symmetry Detection
Date Issued
2011
Date
2011
Author(s)
Lok-Teng Sio, Jolly
Abstract
To perceive a symmetric pattern, an observer needs to find correspondence between two image elements across the symmetric axis. It implies a correspondence relationship between perceptual mechanisms responding to these elements. To perceive a 3D structure in random dot stereogram (RDS), it is shown that perceptual mechanisms responding to different disparities would inhibit each other; and would excite the neighborhoods where the disparity varies smoothly except for a small fraction at the object boundaries. We thus investigated whether putting corresponding elements of a symmetric pattern at different depths would affect symmetry detection. The symmetry patterns consisted of dots (0.19o × 0.19o) occupying 0.5% density of the display. We measured the coherence threshold at 75% accuracy with 2AFC paradigm and manipulated the symmetric pattern rendered on different 3D configuration structures that were produced by the disparity (max 0.228o) between the left and the right eye images. The results show that the two-half plane had inhibited observer’s performance. The coherence threshold for symmetric pattern on a slant surface was similar to that on a frontoparallel plane, even though the depth of the two sides of the symmetric axis was different. The threshold increased dramatically when one side of the axis inclined toward the observer while the other side inclined away even though the depth difference between the two sides was the same as that in the slant condition. The threshold reduced on a hinge configuration whose joint coincide with the symmetry axis. Our result suggests that the co–planarity and the consistency of 2D symmetry pattern and 3D structure are important factors for symmetry detection.
Subjects
symmetry
configuration
surface
depth
slant
shape
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-100-R98227101-1.pdf
Size
23.54 KB
Format
Adobe PDF
Checksum
(MD5):50e6b9ea560c79e2dd4198d2441cc78a