Symmetry Detecting Symmetry in a Random Background Monica Cook
Bilateral Symmetry Bilateral symmetry occurs in many objects and scenes. Eiffel Tower Brooklyn Bridge Animals Faces A Symmetric Target from our Experiment
Bilateral Symmetry Symmetry could be important because: Almost everything is symmetric It directs or catches attention It is important in object recognition and memory
Purpose Hypothesis: Attention and fixation are drawn to symmetric areas. Determine if symmetry can be detected quickly within a random background
Methods Experiment 1: Target located along the horizontal meridian
Methods Experiment 2: Target located at any location within the display
ASL 504 Remote Eye Tracker Participant positioned in a chin rest facing monitor A video camera captures the eye Pupil and corneal reflection used to determine eye position 17-point calibration target Eye camera and infrared LED
Calibration Check A participant looking through the 17-point calibration target after being calibrated.
Example Trial After a video has been calibrated, it looks like this…
Experiment 1: Results Time to First FixationTime to Response Time (sec) Fixation on Symmetric Target
Conclusions: Experiment 1 Symmetry sometimes detected at a glance Detection sometimes required a much longer search Fixated on the target but still hesitated to respond The further the target was from the center, the longer it took to be detected
Experiment 2: Results Average NUMBER OF FIXATIONS
Experiment 2: Results If fixation was drawn to symmetrical areas, this graph would slope progressively downward until the target was found. The blue line below is a good example. DISTANCE IN PIXELS FIXATION NUMBER
Experiment 2: Results However, graphs like these do not follow this pattern.
Conclusions: Experiment 2 The closer the target is to the center, the more quickly it is found Occasionally, participants gradually fixated closer to the target - inconsistent
Why? There may still be preconscious symmetry detection – Accidental symmetry attracts fixations
Further Research A symmetry detection algorithm has been developed to test this. Symmetrical axis of target
The Next Step… Compare fixation locations with the algorithm output for each image If fixations line up with lighter areas: –Indicates eye movements guided by symmetry –Explains why noise in the background is a problem
Thanks ! Jeff Pelz and Andy Herbert Joe Pow and Bob Callens Sue, Brian, Meredith, and Chris The Other High School Interns