Ch 4 - Recognition Flashcards
Knowledge of where our limbs are relative to each other
Proprioception
Knowledge of where our limbs are
Kinesthesis
Information from touch
Haptic information
Ppl use exploratory procedures if trying to identify things by touch
Lederman and klatzky
Simplest form of 2D object recognition
Template matching
Second simplest form of 2D object recognition (and example)
Feature recognition ; pandemonium
What is needed for 2D object recognition to work
Structural descriptions
Marr: object is represented in simplified framework that is broadly the same shape as the object. What’s it called?
Canonical coordinate frame
The building blocks of Marr’s simplified representations
Generalised cones
The first step in creating representation of an object?
Finding a central axis
What has to be done to the object image to get the central axis
Use contour generator to find occluding contours (I.e.silhouette)
Marr’s name for a silhouette
Occluding contours
(Marr ) Silhouette isn’t enough on its own for definite identification, so the way we perceive it is constrained in 3 ways. What are they?
1) each pt on contour generator is a diff pt on the object
2) 2 points that are close on the generator are close on the object
3) all points on the contour generator are on a plane
What’s the problem with the assumption that all points on the contour generator are on a plane?
It would mean that, on a silhouette of a cube, it would look like very different points were on a plane together ( as the silhouette would be a hexagon)
, but this shape is not actually confusing for us
Marr n Nishihara: how do we break up the occluding contours into component parts?
Areas of sharp concavity
