Green Theorems and Qualitative Properties of the Optical Flow

Abstract

The optical flow is the field of apparent velocities derived from the time-dependent image. On the basis of computational arguments it has been suggested that measurements of the optical flow may be used not for a quantitative estimate of physical 3D velocities for which the optical flow is illsuited, but rather for extracting qualitative properties, such as expansion or rotation. These qualitative properties relate directly to physically meaningful properties of the motion and of the 3D structure of the environment: for instance, expansion may signify a fast approach to a surface, that is an imminent crash, while detection of rotation and of its sign may be used to control egomotion. Detectors that could discriminate between translation, expansion and rotation of the optical flow could be quite useful for several visual tasks, such as, for instance, navigation.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 1991
Accession Number
ADA241137

Entities

People

  • Alessandro Verri
  • Tomaso Poggio
  • Vincent Torre

Organizations

  • Massachusetts Institute of Technology

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Artificial Intelligence
  • Artificial Intelligence Computing
  • Boundaries
  • Collision Avoidance
  • Computations
  • Computer Vision
  • Detection
  • Detectors
  • Environment
  • Flow Fields
  • Image Segmentation
  • Information Processing
  • Motion Detectors
  • Navigation
  • Standards
  • Translations

Fields of Study

  • Physics

Readers

  • Calculus or Mathematical Analysis
  • Computer Vision.
  • Educational Psychology