Textbook Notes (373,083)
CA (164,536)
UTSC (18,730)
Psychology (9,797)
PSYB51H3 (315)
Chapter 4

psyb51 chapter 4.pdf

6 Pages

Course Code
Matthias Niemeier

This preview shows pages 1 and half of page 2. Sign up to view the full 6 pages of the document.
Chapter 4 – Perceiving and Recognizing Objects  Middle (midlevel) vision: loosely defined stage of visual processing that comes after basic features have been extracted from image (early vision) and before object recognition and scene understanding (high level vision)  Act of recognition must involve matching what we perceive now to a memory of something we perceive in past Middle Vision  Goal of middle vision is to organize elements of visual scene into groups that we can then recognize as objects  Finding Edges o Because objects abut and overlap other objects, simple connectedness won’t work o Occasional lack of edge doesn’t seem to bother visual system at all o Kanizsa figure – named after Gaetano Kanizsa an Italian psychologist who spent many years investigating stimuli o Illusory contours: contour that is perceived, even though nothing changes from one side of contour to other in image  Rules That Make Contours o Tendency of visual system to go beyond info given was problematic for one of earliest psychologists – structuralists o Structuralism: school of thought believing that complex objects or perceptions could be understood by analysis of components o Structuralists like Wundt argued that perceptions are sum of atoms of sensation – its of color, orientation and so forth  Perception is built up of local sensations the way crystal might be built of array of atoms o Illusory contour challenges view because extended edge is seen bridging gap where no local atom of “edgeness can be found o Gestalt: in German, literally “form”. In perception, name of school of thought stressing that perceptual whole could be greater than apparent sum of parts o Most enduring contribution of Gestalt was to begin description of set of organizing principles  Gestalt grouping rules: set of rules describing which elements in an image will appear to group together. Original list was assembled by members of Gestalt school of thought o Good continuation: Gestalt grouping rule stating that 2 elements will tend to group together if they seem to lie on same contour  Perceptual “Committees” o “all else being equal” phase o Everyone gets together and voices opinions about how stimulus ought to be understood – opinions collide, and results are somewhat unpredictable o Consensus view almost always quickly emerges and settle on single interpretation of visual scene  Occlusion o Why does an edge suddenly stop? o Guess might be that it stops because something else gets in the way, hiding it from view o Answer that visual system seems to come up with is that there’s another contour occluding vertical line, with occluding edge oriented perpendicularly to occluded edge  Texture Segmentation and Grouping o Connecting little pieces of line segments will get only so far in dividing raw image into objects o Texture segmentation: carving image into regions of common texture properties  Closely related to Gestalt grouping principles o 2 strongest principles  Similarity: Gestalt grouping rule stating that tendency of 2 features to group together will increase as distance between them decreases  Texture grouping can be based on similarity in limited number of features – color, size, orientation, and form  Combinations of features don’t work well  Proximity: Gestalt grouping rule stating that tendency of 2 features to group together will increase as distance between them decreases o Weaker principles  Parallelism: rule for figure ground assignment stating that parallel contours are likely to belong to same figure  Symmetry: rule for figure ground assignment stating that symmetrical regions are more likely to be seen as figure o Common region: Gestalt grouping rule stating 2 features will tend to group together if they appear to be part of same larger region o Connectedness: Gestalt grouping rule stating that 2 items will tend to group together if they’re connected  Camouflage o Same principles that are normally used to help find objects in world can be exploited to hide them o Art f camouflage is art of getting features to group with features of environment so as to persuade observer that features don’t form perceptual group of their own  Perceptual Committees Revisited o Low level visual processes deliver fairly straightforward bits of info about line here and color there o Middle vision behaves like collection of specialists, each with specific area of expertise and individual opinions about what input might mean o Goal is to have single answer emerge out of diversity of opinions o Idea of perception by committee has long history, though specific metaphor takes different forms at different times o Early version was Oliver Selfridge’s Pandemonium model for letter recognition (relatively simple subset of object recognition problem)  Called committee members “demons”  Theologically neutral feature demons found vertical lines, acute angles and so forth  Cognitive demons, one for each letter had ideas about features of letters  Cognitive demon looked at work of feature demons and made noise proportional to evidence for its letter  Decision demon listened to committee of cognitive of cognitive demons and identified letter on basis of loudest yell  Committee Rules: Honor Physics and Avoid Accidents o Ambiguous figure: visual stimulus that gives rise to 2 or more interpretations of its identity or structure o Necker cube: outline that is perceptually bi-stable. Unlike situations with more stimuli, 2 interpretations continually battle for perceptual dominance o Necker cubes and duck-rabbits are really exceptions that prove rule o Every image is, in theory, ambiguous, but perceptual committees almost always agree on single interpretation o Accidental viewpoint: viewing position that produces some regularity in visual image that isn’t present in world o Perceptual committee knows about accidental viewpoints and not to bet on them o Second set of assumptions made by visual system involves implicit understanding of some aspects of physics of world o Saying “implicit” means that we need not be able to verbalize rule to use it  Figure and Ground o Ability to distinguish figures (objects in foreground) from ground (surfaces or objects lying behind figures) is critical step on path from image to object recognition o Governing goal is to determine most likely reality behind mage on retina o Figure round assignment: process of determining that some regions of image belong to foreground object and other regions part of background o What principles are at work in assignment of regions to figure or ground?  Surroundedness: rule for figure ground assignment stating that if one region is entirely surrounded by another, likely that surrounded region is figure  Size: smaller region is likely to be figure  Symmetry: symmetrical region more likely to be seen as figure  Parallelism: regions with parallel contours are more likely to be seen as figure  Extremal edges: figure ground calculations are intended to answer question “is region A in front of region B?”  Relative motion: how surface details move relative to edge can determine which portion of display is foreground figure  Dealing With Occlusion o Objects are rarely kind enough to present themselves in splendid isolation on blank backgrounds o 3D objects often hide parts of themselves o Kellman & Shipley stressed ways that edges can relate across gaps o Relatability: degree to which 2 line segments appear to be part of same contour o Heuristic: mental shortcut o Occlusion committee is apparently willing to accept few missed completions In order to reduce vast number of possible completions we would have to consider if we connected every pair of occluded edges o Nonaccidental features: feature of an object that isn’t dependent on exact (or accidental) viewing position of observer  Parts and Wholes o Global superiority effect: finding in various experiments that properties of whole object take precedence over properties of parts of objects. Consistent with implicit assumption we’ve been making throughout our discussion of middle vision – first goal is to carve retinal image into large scale objects o Hoffman & Richards noted that when one blob is pushed into another, pair of concavities is created in silhouette of resulting 2 part object  Summarizing Middle Vision o Goals of mi
More Less
Unlock Document
Subscribers Only

Only pages 1 and half of page 2 are available for preview. Some parts have been intentionally blurred.

Unlock Document
Subscribers Only
You're Reading a Preview

Unlock to view full version

Unlock Document
Subscribers Only

Log In


Don't have an account?

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.