Abstract
Introduction In Lind et al. (2014) (and a number of previous studies), large continuous perspective changes (≥45°) were shown to allow perception of metric 3D shape using stereo- and monocular motion information. None of the previous studies investigated large perspective changes with either use of stereo- or monocular motion alone or in slant perception. The current study explored the effects of small versus large perspective change on the accuracy of metric slant perception under different visual information conditions. Methods We used random dot anaglyphs to display planar surfaces with different upward slant angles rotating around a vertical axis through the surface center and viewed under stereo, monocular, or combined visual information conditions. Within each visual information condition, there were 5 rotation angles from 25° to 65°. Each rotation angle was crossed with 24 different slant angles, from 27° to 73° with 2° increment. The participants’ task was to adjust a line to match the 3D slant of the surface. Results Linear regression of perceived on actual slant for each rotation angle and visual information condition showed that participants underestimated the slant in the monocular condition compared with the stereo and combined conditions. For both stereo and combined conditions, slant judgments were less accurate for rotation angles less than 45° and accurate after the rotation angle reached 45°. However, increasing rotation angle did not affect performance in the monocular condition. Conclusion The current study extended to slant perception the previous finding that large perspective changes (≥45°) allow accurate perception of metric shape. The results also showed that this result requires stereo-motion information, and is not effective when only monocular motion information is available.
Meeting abstract presented at VSS 2015