Example Opening a door How you grasp a handle action is determined by your

Example opening a door how you grasp a handle action

This preview shows page 3 - 5 out of 5 pages.

Example: Opening a door How you grasp a handle (action) is determined by your perception of it Object Recognition Theories Bottom up processing Perception involves processing basic visual elements of an input that are added up for object recognition Passive bottom up processing 1. Gibson’s view Active bottom up processing 1. We use basic elements from visual input to create an internal representation in our mind (involves memory) Pattern recognition theories Identifying something (a pattern) as an instance of a category Find a pattern in visual input and match it to existing patterns 1. Like scanning a bar code Both environment information & existing knowledge direct perception 1. We actively construct perceptions 2. We must recognize configurations from the environment An emerging percept (visual input) has to mingle with a memory trace for perception to occur 1. Template matching 2. Prototype Template Matching theory Match the pattern you see to templates of previous objects you have encountered that you hold in memory A match leads to object identification By this view, how can we recognize NEW objects that don’t match a template 1. Not so flexible (chairs) Prototype theory 1. Compare the configuration of the current visual input with a standard configuration we have in memory (prototype) Prototype: The average representation of the object in memory An idealized version of an object 2. Good enough match = object recognition Hintzman’s multiple-trace memory model A probe (input) in primary memory activates secondary memory traces These traces are activated to the extent that they’re similar to the probe The activated memory traces return an ‘echo’ to primary memory – the general impression of the ‘echos’ determine what we see Feature detection theories Assumes we decompose what we see into a set of features Feature 1. A simple element that can appear in combination with other visual elements to form objects 2. Line, shape Selfridge’s (1959) pandemonium model Basis for connectionist approaches to perception Selfridge’s pandemonium model Bottom level: data or feature demons 1. Individual features are represented Middle level: cognitive demons 1. Detects particular patterns of features 2. If detected the demon ‘shouts’ Top level: decision demons 1. Selects the ‘loudest’ cognitive demon to identify an object Important of feature detection Similar to the primary visual cortex which has special cells/areas for processing different visual features (color, lines, etc.)
Image of page 3
Parallel processing: Performs many computations at once rather than in serial order Recognition by components Combination of feature theory and prototype theory All objects are reducible to a set of basic fundamental geometric shapes called geons 1. 36 basic three-dimensional shapes that can be combined to form any object Recognizing an object involves 1.
Image of page 4
Image of page 5

You've reached the end of your free preview.

Want to read all 5 pages?

  • Winter '08
  • LEVITIN

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture