Visual Processing of Object Velocity and Acceleration

Abstract

Several models of human motion discrimination have been proposed over the last decade. All are loosely related to the correlator approach originally used by Hassenstein and Reichardt to describe motion discrimination in insect eyes. These correlator models use simple spatial and temporal filtering followed by a non-linear multiplicative operation to account for human direction discrimination of sinusoidal stimuli near contrast threshold. Nevertheless, these models are extremely limited in scope. Research in this laboratory has shown that the correlator models cannot explain human speed discrimination even for sinusoidal targets. Apparently, a higher order network formed by combining local 'motion energy units' is required to encode speed. This laboratory has also studied how different features (object components) are combined so that complicated objects move at a uniform velocity. The combination rules are fairly arbitrary, but are limited by physical constraints. Basically, features with similar contrasts, wavelengths, spatial frequencies and temporal frequencies are combined to form a coherent whole moving at a single perceived velocity. Dissimilar features move independently. (aw)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 14, 1989
Accession Number
ADA216509

Entities

People

  • Shawn Mckee

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Abstracts
  • Air Force
  • Availability
  • Computer Vision
  • Contrast
  • Correlators
  • Discrimination
  • Frequency
  • Identification
  • Image Processing
  • Image Recognition
  • Motor Skills
  • Observers
  • Perception
  • Recognition
  • Security

Fields of Study

  • Biology

Readers

  • Radar Systems Engineering.
  • Systems Analysis and Design
  • Vision Science/Vision Psychology/Cognitive Neuroscience.