Abstract:

Purpose: Identifying tilted objects is harder than identifying upright objects using image structure information . However, motion generates optic flow that calibrates image structure. Calibrated image structure preserves spatial layout despite image distortion . Thus, if the tilting process and the tilting objects are continuously perceptible, then the interaction of optic flow and image structure should allow accurate perception of object locations despite orientation change.
Methods and results: We tested this hypothesis by simulating two surfaces separated in depth, with cutout holes in the front surface through which targets and distracters on the rear surface were visible. Both surfaces first rotated and processed rigidly to review depth structures (structure-from-motion), then the front surface translated to occlude targets. Next, the two surfaces tilted rigidly. After some time delays, participants identified locations of hidden targets. We found that: (1) when tilting was perceptible, participants identified object locations equally well with or without orientation change; (2) when tilting process was imperceptible, identifying object locations from a tilted scene after orientation change worsened; and (3) when observers did not see but were told about the amount of tilting, identification was worse than with no orientation change.
Conclusion: Therefore, combined optic flow and image structure information, not mere knowledge about the tilt, enables accurate perception despite orientation change.