Task 5: Mechanisms of Middle- and High- Level Vision Flashcards
Middle vision
loosely defined stage of visual processing that comes after basic features have been extracted from image (low level vision) and before object recognition and scene understanding (high level vision)
What is the goal of middle vision
to organize elements of visual scene into groups that we can recognize as objects
Perceptual organization
the process by which elements in environment become perceptually grouped to create our perception of objects
Name and explain the 2 components of perceptual organization
- Grouping: process by which visual events are put together into units or objects
- Segregation: process of separating one area or object from another
Structuralism
approach that believes that complex objects or perceptions could be understood by analysis of components
Why did Gestalt psychologists reject structuralism? (x2)
- > did not accept idea that perceptions were formed by “adding up” sensation
- > did not accept that past experience plays major role in perception
Name 2 demonstrations that oppose structuralists
- Apparent movement
2. Illusory contours
Discuss the apparent movement principle
principle behind illusion of movement created by stroboscope, in which nothing is actually moving, although movement is perceived
- > 2 images flashing on/off separated by period of darkness which is not seen because our perceptual system adds perception of image moving through space between flashing image
- > whole is different than sum of its parts
Discuss illusory contours
contour that is perceived even though nothing changes from one side of it to the other in an image
- > edges create triangle despite no physical edges being present
- > cannot be created y sensations
- > whole is different than sum of its parts
How do the Gestalt principles act like committees?
some overrule, some don’t
Discuss law of pragnanz/good figure/simplicity
every stimulus pattern is seen in such a way that the resulting structure is as simple as possible e.g. olympic symbol
Discuss law of similarity
similar things appear to be grouped together due to colour, shape or orientation
Discuss law of good continuation
points that when connected result in straight or smoothly curving lines are seen as belonging together and the lines tend to be seen in such a way as to follow the smoothest path
Discuss law of proximity
things near each other appear to be grouped together
Discuss principle of common region
elements that are within the same region of space appear to be grouped together
What does the principle of common region override?
law of proximity
Discuss the principle of uniform connectedness
connected region of visual properties e.g. lightness, colour, texture is perceived as a single unit
Discuss the principle of common fate
things that are moving in same direction appear to be grouped together
Discuss the principle of closure
applies when we tend to see complete figures even when part of the information is missing
Differentiate between figure and ground
ground - background
figure - seperate object standing out from ground
Define reversible figure-ground
patterns that can be perceived different depending on what we see as figure or ground
Define border ownership
even if figure and ground share contour, border is associated with figure
Name 4 factors determining which area is figure
- area
- symmetry
- meaningfulness
- roundedness
How does area determine which area is figure?
lower region of a display tends to be seen as figure, there is no left-right preference
How does symmetry determine which area is figure?
symmetrical areas/convex regions are seen as figures
How does meaningfulness determine which area is figure?
displays that look familiar to us are more likely to be seen as figures
How does roundedness determine which area is figure?
a rule for figure-ground assignment stating that if one region is entirely surronded by another, it is likely that the surronded region is the figure
Gist of a scene
meaning of a scene that can be recognized after only a fraction of a second
Global image features
information used by us, that can be perceived rapidly and is associated with specific types of scenes, contains info on scene’s structure and spatial layout
Name 5 global image features
- degree of naturalness
- degree of openness
- degree of roughness
- degree of expansion
- colour
Discuss how degree of naturalness is a global image feature
– Natural scenes (= ocean) have undulating contours while man-made scenes (= street) have straight lines, like horizontals and verticals
Discuss how degree of openness is a global image feature
Open scenes (= ocean, street less) have visible horizon line and few objects, while scenes with low degree of openness (= forest) do not.
Discuss how degree of roughness is a global image feature
Low roughness scenes (= ocean) have few small elements, while high roughness ones (= forest) have small and complex elements
Discuss how degree of expansion is a global image feature
High degree of expansion means that there is convergence of parallel lines (think about railroad tracks in distance).
-> dependent on observer’s viewpoint
Discuss how colour is a global image feature
scenes have characteristic colours, ocean scene is blue, forest green/brown etc
Regularities in environment
a) define
b) name 2
a) Characteristics of environment that occur frequently, such as landscapes are often green and smooth, and verticals and horizontals are associated with buildings.
b) physical and semantic
Physical regularities
regularly occurring physical properties of the environment
- > more verticals+horizontals in environment than oblique explaining oblique effect
- > objects have homogenous colours
Light-from-above assumption
– Assumption that light is coming from above. Made because most light in our environment actually comes from above.
Semantic regularities
Characteristics associated with functions carried out in different types of scenes.
-> meaning of a scene
How do people infer what is present in a scene?
use physical and semantic regularities
Discuss retinal ambiguity in relation to Helmholtz
particular pattern of stimulation on retina that can be caused by many different possible objects in environment
Discuss theory of unconscious inference in relation to Helmholtz
some of our perceptions are the result of unconscious assumptions we make about the environment.
Discuss the likelihood principle in relation to theory of unconscious inference
we perceive object that is most likely to have caused pattern of stimuli we have received
Nonaccidental feature
of an object is not dependent on the exact viewing position of the observer when occluded
Global superiority effect
the finding in various experiments that the properties of the whole object take precendence over the properites of parts of the object
Bayesian inference
Way of formalizing idea that our perception is combination of current stimulus and our knowledge about conditions of the world – what is and is not likely to occur.
-> Statistical technique that enables us to calculate the probability (P) that the world is in a particular state (A) given a particular observation (O).
Link the visual areas associated with each level of vision and describe
V1 – Low-level vision – Cells respond best to lines and edges in specific areas of visual field.
V2 – Middle-level vision – Cells have sensitivity to border ownership and illusory contours.
V4 – High-level vision – Cells are more interested in complex attributes. No perfect stimuli have been found for this area, but warmer and darker colors generate more response, as well as stimuli with feature pointing to the right.
The ___ the response the more likely someone can identify something
larger
Binocular rivalry
if each eyes receives totally different images, the brain can’t fuse the 2 images therefore the observer perceives either the left-eye image or right-eye image (but not both at the same time)
Naive template theory
Proposal that visual system recognizes objects by matching the neural representation of the image, with a stored representation of the same ‘shape’ in the brain.
Discuss lock and key in relation to the naive template theory
Stimulus has to fall on array of spot detectors. If it falls on filled detectors and not empty ones, it is identified.
Recognition-by-components model
Biederman’s model of object recognition which holds that object’s are recognized by identities and relationships of their component parts (based on Gestalt principles: symmetry, parallelism etc)
Geon
Any of the ‘geometric ions’ out of which perceptual objects are built.
Viewpoint invariant
Property of object proposing that representations of object do not change when observer viewpoint changes.
Entry-level category
- subordinate level
- superordinate level
For object, label that comes to mind most quickly when we identify it (= bird).
- Subordinate level – Object might be more specifically named (= eagle).
- Superordinate level – Object might be more generally named (= animal).
Pure alexia
inability to recognize a single word, faces, objects, digits, numerals - can speak and even write
Neuronal recycling hypothesis
– Human brain contains highly organized cortical maps that constrain subsequent learning.
Where do illusory contours evoke response?
area 18 of VC (V2)
Population coding
information about the outside world and plans for motor action are represented by the firing patterns of neurons – temporal sequences of action potentials, also known as spikes
Population codes
are neural representations at the level of groups of cells.