Final Flashcards
How do distance and depth bring back the inverse problem?
How does the visual system recover information about the 3D spacial structure of the environment if it has to get information about an object’s distance and orientation of the surface in order to do so?
Visual system must compute the orientation and distance of each surface reflecting light to the retina.
What are the pieces of information necessary for orientation and distance of surfaces?
- Slant
- Distance
Slant
angle between line of sight to the surface patch and its normal (the direction perpendicular to the surface)
Tilt
direction of a surface’s depth gradient (orientation)
Any stimulus is ambiguous (can come from distal point at any distance), so how does visual system determine depth and distance of objects in visual field?
(among other things)
- accomodation
- convergence
- binocular disparity
- motion parallax
What is accomodation?
- ocular, monocular, absolute D info
- lens shape changes when fixated object changes in distance.
- lens is flat at greater distances and curved at short distances, which serves to focus image
- Focal point: the point behind the lens where light rays passing through it converge.
- If accomodation fails, image is blurry. Elderly have accomodation problems because the elasticity of the lense decreases with age.
What is convergence?
- ocular, binocular, absolute
- the angle formed by the direction of gaze of two eyes when both are fixating on an object.
- The closer the object, the larger the convergence angle. The furhter away the object, the smaller the convergence angle.
- Reduces double images of objects.
What is binocular disparity?
Disparity that occurs because our eyes are a few inches apart, so things are viewed from two different distances in space and there is an imperfect overlap.
- Visual system disparity into depth, so it is a relative depth cue.
- Direction of disparity says what’s in front of what.
What is a horopter?
a notional curved surface that delineates corresponding retinal points
What is a stereoscope/stereoscopic depth?
- stereoscope: contraption that allows each eye to view a different image and allows for artificial depth to be viewed.
- Stereoscopic depth: the disparity that arises from a difference in widths between images transmitted to the eyes, signaling depth.
- We are sensitive to disparity up to about 100 feet.
What is the correspondence problem?
Which element in the right eye’s image correspond to what element in the left eye’s image? Shape may play a role in allowing the visual system to create a correspondence between these two images.
What are the two theories regarding temporal shape analysis?
- Shape first theory: The left image and right image exist, then each eye individually analyzes the shape of the image. Once that shape analysis is completed, stereopsis can occur.
- Stereopsis first theory: The left image and right image exist, then they are fused and stereopsis occurs. Once that happens, shape can be analyzed.
Random Dot Stereogram
- no monocular shapes
- shows that disparity analysis precedes shape analysis since there’s no shape in either eye’s view.
- Thus, shape is derived from and follows disparity analysis and the resolution of the correspondence problem.
What is binocular rivalry?
- grossly mismatched images are projected to each eye, and rivalry occurs, since one image is seen while the other image is suppressed.
- suppressions alternate
Why do we experience singularity of vision, even though we always have two different images on our retinas?
May be:
- alternating suppression of one image by the visual system
- fusion: the images on the retina fuse together to form one cohesive image.
What are the rules governing binocular rivalry?
- figure dominates ground
- figure carries area of ground with it
- meaningful image dominates
- evidence for suppression theory: if you close one eye, light doesn’t appear dimmer, and if you’re looking at a light grey stimulus in one eye and a dark grey stimulus in the other, the percept isn’t medium grey.
What is motion parallax?
- monocular, dynamic, relative to depth
- perceived differences in movement between near and far objects
- caused by changes in the angle of the stimuli on the retina as one moves further across the retina than the far object, which makes closer object appear to move faster than further object.
- Fixated point does not displace
- points closer than fixation displace in the opposite direction
- good for relative depth information at great distances, unlike disparity
Looming
If something on a screen increases in size but maintains its dimensions, it looks like it’s moving toward the observer. Babies show this, as well.
What are the configural (pictorial) depth cues?
- perspective (linear and size)
- texture density gradients
- height in plane
- occlusion
- shadow (cast and attached)
- familiar size
Information that is available from just looking at a picture
What are the ocular cues?
- accomodation
- convergence
- binocular disparity
- motion parallax
- accretion and deletion of texture
Perspective
- linear perspective is strongest source of configurational depth information.
- parallel lines project converging parallel lines at same vanishing point onto the retina
- 3D scenes project a 2D image onto flat surface, but is recoverable from configurational depth cues (inverse problem addressed)
- size perspective: if it’s smaller, it’s further away, like Gibson cylinders
What is the Kinetic Depth Effect?
- the three-dimensional structural form of an object can be perceived when the object is moving
- if a shadow is cast onto a screen by a rotating wire shape, a viewer can readily perceive the shape of the structure behind the screen from the motion and deformation of the shadow.
- Assumption of rigidity: The visual system prefers to see a rigid object rotating than the 2D object morphing.
Height in plane
position relative to the horizon line. The further above it is, the higher it is. Requires perceiving ground plane. Not super powerful, but can be accompanied by another type of perspective.
Texture density gradients
- Gibson
- Density of texture increased with distance (field of flowers)
- The steeper the gradient, the greater the slant in depth, sharp discontinuity gives information about edges.
- direction of gradient gives direction of slant
Foreshortening
circles looking elliptical instead of like circles implies a change in depth or position. Objects appear more oblong or distorted from their true shape.
Aerial Perspective
Things are blurred and bluish at greater distances (mountains in distance). Not as powerful as other types of perspective.
Interposition
- partial occlusion is seen as depth
- the occluding shape says it’s in front, but there’s no information about how far in front it is.
- does not require familiar figures
Two types of shadows
- Cast: person’s shadow reflected on sidewalk
- Attached: shadows from contours on the object itself (noses, indentation of eyes). When light falls on convex object, it casts shadow below it. When light falls on a concave object, it casts shadow above it. Shadow location gives strong information about depth of an object.