Motion Detection at Corners

Abstract

This is the first in a series of reports which describes an approach to image velocity estimation and subsequent moving object extraction and tracking. Although the approach utilizes some of the gradient-based motion estimation techniques mentioned above, it has significant differences from those techniques. This paper is concerned with this first step and describes techniques for measuring image motion at corners. Corners have the property that their motion can be directly computed based only on measurements made at the corner. Another important property of corners is that they can be safely regarded as projections of scene features whose general appearance is invariant to rigid motion - e.g., an image corner may be the projection of the vertex of a polyhedron, or of a curvature discontinuity on the boundary of a surface marking. The second step involves propagating these velocity estimates to a larger number of picture points.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1981
Accession Number
ADA124806

Entities

People

  • Hanfang Sun
  • Larry S. Davis
  • Zhongquan Wu

Organizations

  • University of Maryland

Tags

Communities of Interest

  • C4I

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Boundaries
  • Computations
  • Computer Science
  • Computer Vision
  • Coordinate Systems
  • Detection
  • Detectors
  • Graphics
  • Image Processing
  • Intensity
  • Least Squares Method
  • Mathematics
  • Night Vision
  • Translations
  • Universities

Readers

  • Combustion Dynamics and Shock Wave Physics.
  • Control Systems Engineering.
  • Neural Network Machine Learning.