Neural Network Methodologies and Their Potential Application to Cloud Pattern Recognition.

Abstract

Six artificial neural network models are explored as potential methodologies for the automated interpretation of satellite cloud images. The Multi-layer Perceptron Network is chosen to be the most applicable to the image interpretation problem. A complete, mathematical description of the methodology is presented. The neural network's classification capability is demonstrated using simple geometric patterns and alphabetic characters. A more complex test using GOES infrared imagery shows that the neural network can distinguish 53 of 54 large-scale cloud patterns. An architecture for a complete, automated cloud feature recognition system is proposed.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1991
Accession Number
ADA239214

Entities

People

  • J. E. Peak

Organizations

  • Computer Sciences Corporation

Tags

Communities of Interest

  • Energy and Power Technologies
  • Space

DTIC Thesaurus Topics

  • Algorithms
  • Artificial Intelligence
  • Artificial Intelligence Computing
  • Artificial Intelligence Software
  • Artificial Satellites
  • Automata Theory
  • Classification
  • Cognition
  • Computer Science
  • Computers
  • Expert Systems
  • Grids
  • Identification
  • Machine Learning
  • Neural Networks
  • Pattern Recognition
  • Recognition

Fields of Study

  • Computer science

Readers

  • Atmospheric Remote Sensing.
  • Business Analytics
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks
  • Space