2 - Object Recognition

Question 1

Perceptual organization

Answer 1

-take info from both processing types and make into complete pic

Question 2

Bottom-up processing

Answer 2

• Data driven
• Have no background info/reference
• Take sensory stimuli like sight and combine, create cohesive image, then determine what object is
• build whole thing from parts

Question 3

Top-down processing

Answer 3

Conceptually driven

• start w/ expectation or memory
• recognize whole thing first, then recognize components based on expectations

you can walk through your room in dark w/o hitting anything, maybe slower, b/c you know where things are

Question 4

Optical illusions

Answer 4

• take advantage of perceptual organization

- based on differences in processing

Question 5

Gestalt principles

Answer 5

• when you look at a bunch of objects / complicated images, your brain groups those things to be as non-complex / concise as possible
• group based on similarity in size/shape, etc
• whole = sum of parts

Proximity, Similarity, Good Continuation, Subjective Contours, Closure

(see book, make notes)

Question 6

Proximity

Answer 6

dots arranged in triangle is perceived as a shape, not individual dots
-your brain can count 5 objects or less w/o actually counting them

Question 7

Similarity

Answer 7

Group similar things

ex) big dots in triangle shape against background of tiny dots rather than a bunch of random dots

Question 8

Good continuation

Answer 8

• we see lines as continuous/straight or curved

- we don’t like jagged- too complicated

Question 9

Subjective Contours

Answer 9

Foreground vs background

• foreground is black half-shapes, invisible “square” seems like it instead
• brain “fills in the gap”- illusory contour

Question 10

Closure

Answer 10

4 L shaped things is a square- easier to perceive

Question 11

Law of Pragnanz

Answer 11

Perceptual organization will always be as regular, simple, and symmetric as possible
explains gestalt

Question 12

Depth Perception - Visual cues to deduce distance

Answer 12

Monocular

• require one eye
  1) relative size = larger is closer
  2) interposition = overlapping objects; front is closer
  3) linear perspective = convergence of parallel lines in distance; greater convergence = further distance
  4) motion parallax = closer objects move faster when we change field of vision / look at something else

Binocular cues

1) retinal disparity (2.5 inches apart)
- used in VR- give each eye slight dif image
2) convergence- brain detects angle b/w 2 eyes required to bring object into focus
- distance object = both eyes stare straight ahead; looking at nose = extreme angle
- used in parvocellular cells and feature detection, and motion through magnocellular cells
3) constancy - some objects remain same despite environment changes- color, brightness, size, shape