Chapter 6: Flashcards
How do we convert 2-D to 3-D?
- egocentric distance (how far is something from us)
- relative distance (how far is something from another object?)
Oculomotor Cues (2)
- Accomodation (lens change for close ob)
- Convergence (eyes converge as ob gets closer to eyes)
- both for egocentric, reaching/grasping
Monocular Cues: Static (pictorial)
- The arrangement and appearance of objects in a scene is partially determined by their location in space
- Pictorial cues are named for the fact that they can be represented in pictures to give realistic appearance
- Early paintings did not recognise the importance of this information
- Da Vinci one of the first to use perspective systematically
- These cues are monocular because they can be recognised with one eye
Static Cues: 3 groups, 9 altogether
Position-Based Cues --partial occlusion (V2) --relative height Size-Based Cues --familiar size --relative size --texture gradients --linear perspective Lighting-Based Cues --atmospheric perspective --shading --case shadows
Relative Height
relative to horizon/point of fixation
- how far from horizon
- further away from point, the closer it is to us
Size and Distance
- farther objects looks smaller on retinal field, closer is larger.
- if put out two ballots and then pump one up, it looks like its rushing forward b/c only have retinal field info (removed depth)
Texture gradients:
Texture elements get closer together and smaller with increasing distance
Texture discontinuities:
Similar to occlusion; texture changes abruptly at depth boundaries
Atmospheric (or aerial) perspective
more atmo=more air=more things to scatter light
-therefore, fugier, more blue, less kontrast, less distinct edges
Monocular Cues: Dynamic (3)
motion parallax
optic flow
deletion and accretion
Motion parallax:
- position of ob changes as we move through a scene
- Objects that are closer move past the observer more quickly and appear to move in a different direction than more distant objects
- Very strong cue
- Provides accurate quantitative information about distance
- closer moved larger distance on retina (one side onto the other), vs. far object move on a little bit (stayed on same side as fixation)
Binocular cues
- Binocular cues are based on the fact that we have two forward facing eyes that are laterally separated
- This provides slightly displaced images in each eye
- This information can be converted into a signal about relative depth
- Based on the geometry of the images reaching the eye
Binocular Disparity
- ob closer=crossed disparity (close ob on right side of left eye)
- ob farther=uncrossed disparity (far ob on left side of left eye)
- Each distance will produce a different amount of retinal disparity
- Binocular cells demonstrate that cells are responsive to varying amounts of binocular disparity
Horopter
- fixation plane
- When fixating, image of target falls on fovea of each eye
- The images of an object at the same distance as the fixation plane will fall on the same relative position in the two eyes
- Images that fall on different relative locations are said to fall on non-corresponding (disparate) points
Fixation point
zero retinal disparity
Stereopsis
the ability to use binocular disparity as a cue to see depth
-how do we match stuff?
Local Vs. Global Stereopsis
Local: [With simple objects, matching could be done on a point by point basis, would require object or feature recognition prior to stereopsis]
Global: [whether retinal disparity was a sufficient cue for stereopsis in the absence of any other depth cues
Created random-dot stereograms – no local stereopsis (no objects or contours) or static monocular depth cues]
Size Constancy
- Many objects at different distances could result in the same size retinal image
- problem of size constancy is how we have stability of perceived size despite these large changes in the size of the retinal image
- again, perception seems to be linked to the invariant characteristics of objects (distal stimuli) rather than the enormously changing proximal stimuli
Emmett’s Law (size-Distance Invariance)
Perceived Size = k*(Retinal image size x Perceived Distance)
-where you see ob correlated to the size of the afterimage
The Ponzo Illusion
- Car closer looks smaller then far car but both are same size
- The two spheres are the same size but the perceptually larger one evoked activity in a larger area of V1.
Ames Room
(a) This shows the actual shape of the Ames room. Most of the visible surfaces of an Ames room are trapezoidal, but they all appear rectangular to a person looking through the peephole. Thus, the left and right corners of the wall opposite the peephole appear to be equidistant from the observer, and the wall appears to be perpendicular to the line of sight.
(b) In this photo taken through the peephole in an Ames room, the children on the right look taller than the adult woman on the left, but of course they are much shorter.
Moon Illusion
- Moon (or sun) seems larger at the horizon than at the zenith
- Recognised in classical times, many theories
- Current most-accepted explanation in terms of apparent distance, although issue is still controversial
- Assumption: if two objects have the same retinal image size, the one that appears closer will look smaller
- That means horizon moon must look more distant
- Some evidence that horizon looks further away than zenith sky
