Keywords
Binocular vision; Stereopsis; Inverse optics; Multiple view geometry; Inverse projective geometry;
Abstract
The human visual system uses binocular disparity to perceive depth within 3D scenes. It is commonly assumed that the visual system needs oculomotor information about the relative orientation of the two eyes to perceive depth on the basis of binocular disparity. The necessary oculomotor information can be obtained from an efference copy of the oculomotor signals, or from a 2D distribution of the vertical disparity, specifically, from the vertical component of binocular disparity. It is known that oculomotor information from the efference copy and from the vertical disparity distribution can affect the perception of depth based on binocular disparity. But, these effects are too slow and too unreliable to explain the stable and reliable depth perception we have under natural viewing conditions when natural eye movements are made. This study describes a computational model that recovers depth from a stereo-pair of retinal images without being given any oculomotor information.
Start Date
15-5-2019 2:30 PM
End Date
15-5-2019 3:00 PM
Included in
Cognition and Perception Commons, Cognitive Psychology Commons, Theory and Algorithms Commons
Recovering Depth from Stereo without Using Any Oculomotor Information
The human visual system uses binocular disparity to perceive depth within 3D scenes. It is commonly assumed that the visual system needs oculomotor information about the relative orientation of the two eyes to perceive depth on the basis of binocular disparity. The necessary oculomotor information can be obtained from an efference copy of the oculomotor signals, or from a 2D distribution of the vertical disparity, specifically, from the vertical component of binocular disparity. It is known that oculomotor information from the efference copy and from the vertical disparity distribution can affect the perception of depth based on binocular disparity. But, these effects are too slow and too unreliable to explain the stable and reliable depth perception we have under natural viewing conditions when natural eye movements are made. This study describes a computational model that recovers depth from a stereo-pair of retinal images without being given any oculomotor information.