Abstract
Kinetic occlusion (the accretion and deletion of texture of a background region by a solid foreground object) has been shown to be an important cue for determining depth order and discrimination of 2-D shapes. In particular, Andersen and Cortese (1989) showed that kinetic occlusion can be used to discriminate 2-D shapes and that density has a more important role than velocity in determining accuracy. In the present study, we developed a model that uses spatial/temporal integration to identify 2D shape information from kinetic occlusion. The model was tested with 4 opaque 2D shapes (square, circle, star, and diamond) presented in front of random dot texture. We used a Bayesian analysis to select a shape among its' four alternative choices using a distance metric based on the average location of accretion and deletion of image texture. The velocity and density of the background dots was varied similar to study by Andersen and Cortese. Model performance increased with an increase in the velocity of the background texture and with an increase in the density of the background texture, with density resulting in a greater impact on shape discrimination. Parameters for spatial and temporal integration that predict human observer performance will be discussed.
Supported by NIA grant 2R01AG013419-06A2.