Recovering 2-D Motion Parameters in Scenes Containing Multiple Moving Objects.

Abstract

A method for extracting the motion parameters of several independently moving objects from displacments field information is described. The method is based on a generalized Hough transform technique. Some of the problems of this technique are addressed and appropriate solutions are proposed. A modified multipass Hough transform appraoch has been implemented, where i each pass windows are lkocated around objects and the transform is applied only to the displacement vectors contained in these windows. The windows are determined by the degree to which the displacement field is locally inconsistent with previously found motion transformations. Thus, the sensitivity of the Hough transform to local events is increased and the motion parameters of small objects can be detected even in a noisy displacement field. The author also use a multi-resolution scheme in both the image plane and the parameter space and thus reduce the computational cost of the technique. The method is demonstrated by experiments based on artifical images with four parameters of 2-d motion: rotation, expansion and tranlation in both axes.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 01, 1983
Accession Number
ADA129565

Entities

People

  • Gilad Adiv

Organizations

  • University of Massachusetts Amherst

Tags

Communities of Interest

  • Air Platforms
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Boundaries
  • Computations
  • Computer Graphics
  • Computer Science
  • Computer Vision
  • Computers
  • Coordinate Systems
  • Detection
  • Image Processing
  • Pattern Recognition
  • Reliability
  • Rotation
  • Three Dimensional
  • Translations
  • Two Dimensional

Readers

  • Computer Vision.
  • Control Systems Engineering.

Technology Areas

  • Space
  • Space - Space Objects