EPBST Workstation Improvement: Automated Point Matching

Abstract

As part of the workflow of the EPBST workstation software for building terrain models, point correspondences (x1, y1), (x2, y2), representing the locations of a common physical feature in two images, must be identified. Early versions of the software required that all such point correspondences be identified manually. This report describes the methodology used in adding automated point matching to the software, using: (1) interest point detection, (2) correlation, (3) consistency checks, and (4) subpixel refinement by least squares adjustment.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 11, 2001
Accession Number
ADA397084

Entities

People

  • Stephen J. Dow

Organizations

  • University of Alabama in Huntsville

Tags

Communities of Interest

  • Materials and Manufacturing Processes
  • Weapons Technologies

DTIC Thesaurus Topics

  • Algorithms
  • Army Aviation
  • Cameras
  • Computations
  • Consistency
  • Detection
  • Geometry
  • Governments
  • Grids
  • Image Registration
  • Images
  • Numbers
  • Real Numbers
  • Remote Sensing
  • Terrain Models
  • Two Dimensional
  • Visible Spectra

Fields of Study

  • Computer science
  • Engineering

Readers

  • Computer Vision.
  • Software Engineering.