Terrain Analysis Using Landsat Thematic Mapper Imagery

Abstract

This thesis examined two sites from a Landsat scene of portions of Honduras and Nicaragua. One site was examined for potential water obstacles, and the other was examined for cover and concealment provided by vegetation. The results suggest that potential water obstacles can be detected. It is not clear if vegetative cover and concealment can be reliably detected. A study using better ground reference information than was available is necessary to answer that question. Several unsupervised classification algorithms were used and compared. A histogram clustering algorithm followed by a minimum distance classifier provided results comparable to the much slower K-means and isodatatype algorithms. Several methods to reduce the dimensionality of the classification problem were examined, including band subsets, between-band ratios, the principal component transformation, and the tasseled cap transformations. Band subsets provided adequate accuracy and is the easiest method to implement. (edc)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1990
Accession Number
ADA225704

Entities

People

  • Gerald T. Michael

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Energy and Power Technologies
  • Sensors
  • Space
  • Weapons Technologies

DTIC Thesaurus Topics

  • Algorithms
  • Computers
  • Detection
  • Detectors
  • Engineering
  • Geography
  • Identification
  • Image Processing
  • Information Science
  • Machine Learning
  • Military Operations
  • Pattern Recognition
  • Recognition
  • Remote Sensing
  • Supervised Machine Learning
  • United States
  • Unsupervised Machine Learning

Readers

  • Applied Combinatorial Optimization and Logic Circuit Design.
  • Computer Vision.
  • Environmental Remediation and Restoration.