Image Interpretation Using an Expert System

Abstract

The purpose of this thesis is to study and enhance an automatic image feature analysis system for aerial photographs in order to detect changes with respect to a model defined in a knowledge-base. The system interpretation must resemble some intelligent human interpretation and by its use should be able to reduce the manual effort in processing large volumes of data. This computerized rule-based system is integrated by using Prolog and Image processing operators, which run in a multiprocess environment and communicate through a blackboard storage. The Prolog expert provides the model through a collection of certainty rules and facts, and uses its inference engine capability to combine different measurements. They are obtained as the evidence necessary to deduce a conclusion about the condition of a feature. The system must be able to deal with uncertainties generated by noise in the images, variability of imaging conditions, and possible errors in the model. The main task is the formulation of interpretation rules so that the expert system can mimic the reasoning effectively from the domain principles and yield high confidence results. Keywords: photogrammetry; aerial image interpretation; Fortran.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1986
Accession Number
ADA176911

Entities

People

  • Diego L. Rueda

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • C4I
  • Space

DTIC Thesaurus Topics

  • Aerial Photographs
  • Artificial Intelligence
  • Artificial Intelligence Software
  • Cameras
  • Computer Languages
  • Computer Programming
  • Computer Programs
  • Computer Vision
  • Computers
  • Coordinate Systems
  • Databases
  • Expert Systems
  • Image Processing
  • Inference Engines
  • Photographs
  • Three Dimensional
  • Two Dimensional

Readers

  • Distributed Systems and Data Platform Development
  • Geodesy
  • Regression Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference