whereRU: Aresty Poster 177- Illusory Figures, Clustering, and Multiple Object Tracking by Aresty Posters 2009

Illusory Figures, Clustering, and Multiple Object Tracking

Abstract:
In typical Multiple Object Tracking (MOT) tasks, objects are well-defined shapes. We examined whether illusory figures and non-illusory feature matched clusters (constructed of the same parts as the illusory figures but oriented so that no illusory figure is seen) can be tracked as individual objects in MOT. We presented illusory Varin ring triangles, non-illusory Varin ring clusters, and Varin ring triangles with drawn contours and measured both tracking performance and probe-dot detection on target and nontarget objects. Both measures support the hypothesis that illusory figures and non-illusory clusters are tracked as individual objects: figures with illusory contours, feature matched clusters and figures with drawn contours yielded the same pattern of results. Probe-dot detection on nontargets was significantly worse than detection on targets for all conditions. Tracking performance exceeded 86% for all figures.

Author: Ryan McKendrick (rmckz@eden.rutgers.edu)
Advisor: Zenon Pylyshyn (zenon@ruccs.rutgers.edu)
Lab Hompage: ruccs.rutgers.edu/finstlab/finstsum.html