Determining the Location of an Observer with Respect to Aerial Photographs

Abstract

In this study, the possibility of using a computer to detect the motion of an observer by comparing two successive aerial photographs is examined. The purpose of the study was to experiment with a technique for finding a point common to both images. The technique presented uses only sharp boundary lines and their distribution in the images to produce the 'primal sketches' of the image. Once the 'primal sketches' are made, the original images are not referred to again. A point common to both images is identified by comparing the number of cells with strong gradient magnitudes and their distribution in 3-pixel by 3-pixel blocks. The technique produces excellent results in analyzing simulated successive photographs, suggesting good results with photographs that are taken in succession by a moving observer. Compared with other work on image correlation and object identification, the technique uses fewer features (only two) in its primal sketches, and it does not need any human intervention. Possible applications are photo interpretation, high-altitude navigation, and underwater station-keeping. Keywords: Theses

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1988
Accession Number
ADA207791

Entities

People

  • Jill D. Wolfe

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Air Platforms
  • Autonomy
  • Engineered Resilient Systems

DTIC Thesaurus Topics

  • Aerial Photographs
  • Arrays
  • Artificial Intelligence
  • Classification
  • Colors
  • Computer Science
  • Computer Vision
  • Computers
  • Identification
  • Images
  • Lisp Programming Language
  • Navigation
  • Pattern Recognition
  • Seabed
  • Security
  • Statistical Analysis
  • United States

Readers

  • Computer Vision.