AN APPLICATION OF INTEGRAL GEOMETRY OF PATTERN RECOGNITION

Abstract

A machine for recognizing patterns may be considered to consist of two principal parts--a receptor and a categorizer. The receptor views the pattern and translates its features into machine language. The categorizer classifies the pattern using the information supplied to it by the receptor. In this paper information extracted from a visual pattern by receptor is required to be invariant with respect to the rotation and the translation (and at times, scale changes) of the pattern. We thereby obtain numerical parameters from the pattern that correspond to the shape of the pattern. Integral geometry is employed to obtain these invariant parameters. Probabilities of error are found as a function of the number of pattern measurements which are used to estimate a parameter. The numerical values of some parameters are obtained for the set of patterns consisting of circles, squares, rectangles, ellipses and right isosceles triangles. In addition, decision theory is used to obtain the structure of the test having minimum probability of error. (Author)

Document Details

Document Type
Technical Report
Publication Date
Feb 01, 1962
Accession Number
AD0275352

Entities

People

  • G.h. Ball

Organizations

  • SRI International

Tags

DTIC Thesaurus Topics

  • Decision Theory
  • Geometry
  • Integrals
  • Language
  • Machine Languages
  • Mathematics
  • Measurement
  • Pattern Recognition
  • Probability
  • Recognition
  • Rotation
  • Translations
  • Triangles

Readers

  • Computer Vision.
  • Electromagnetic Wave Scattering and Antenna Radiation Engineering

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference