Abstract/Details

A rotation-invariant image pattern classifier


1990 1990

Other formats: Order a copy

Abstract (summary)

A pattern recognition system is presented that is able to locate and classify image patterns in aerial photographs without a priori knowledge of the pattern position and orientation. The system contains three basic parts: a feature extractor, a screening detector, and a classifier. The feature extractor is based on spatial correlation using filters derived from the Circular Harmonic Transform (CHT) of a set of reference patterns. The detector is a one-pass filter used to detect targets and the classifier assigns a target class to the patterns found by the detector.

The performance of the system is estimated using Monte Carlo analysis. Experimental results indicate that the system is capable of locating and classifying patterns of similar types with a probability of error less than 10%. Estimates of the probability of false alarm are less than 16 $\times$ 10$\sp{-6}$ per pixel. Empirical receiver operator characteristics are also presented.

Indexing (details)


Subject
Electrical engineering;
Computer science
Classification
0544: Electrical engineering
0984: Computer science
Identifier / keyword
Applied sciences; computer vision; pattern recognition
Title
A rotation-invariant image pattern classifier
Author
Budge, Scott E.
Number of pages
167
Publication year
1990
Degree date
1990
School code
0022
Source
DAI-B 51/12, Dissertation Abstracts International
Place of publication
Ann Arbor
Country of publication
United States
Advisor
Chabries, Douglas M.
University/institution
Brigham Young University
University location
United States -- Utah
Degree
Ph.D.
Source type
Dissertations & Theses
Language
English
Document type
Dissertation/Thesis
Dissertation/thesis number
9112808
ProQuest document ID
303802850
Copyright
Database copyright ProQuest LLC; ProQuest does not claim copyright in the individual underlying works.
Document URL
http://search.proquest.com/docview/303802850
Access the complete full text

You can get the full text of this document if it is part of your institution's ProQuest subscription.

Try one of the following:

  • Connect to ProQuest through your library network and search for the document from there.
  • Request the document from your library.
  • Go to the ProQuest login page and enter a ProQuest or My Research username / password.