Contributions of geometric and image information in the perception of solid objects

Abstract

Purpose: At ARVO 2000 we extended the paradigm used by Pelli (1999) for investigating the relationship of image information to the perception of 3-D objects. Our experiments demonstrated an interaction of two dimensional scale and three dimensional structure on our perception of 3-D shape. In this series of experiments we further manipulate the nature of the image information available to the viewer to assess its perceptual utility. Methods: Synthetic objects with a controllable range of 3-D spectra and 2-D image resolution were created to match the experimentally relevant stimuli parameters from our previous work. The resulting stimulus images were then mapped onto an orientable plane. Field of view, 2 and 3-D spatial frequency, and slant (from -35 to 35 degrees) of the plane relative to the viewer were manipulated. Observers indicated if the subject matter appeared to be two or three dimensional on each trial. Results: Results for the frontoparallel stimuli were similar to those found in our previous experiments. Stimuli whose image was slanted relative to the observer showed no significant difference in criterion suggesting that viewers were making their decisions in a more object than image relative space. The criteria was different across observer but consistent within each. Conclusions: These findings suggest that the effects of 2-D image and 3-D geometric structure are largely independent of the orientation of the projected image relative to the observer, at least for the range of slants used in this experiment.