Generalized Cooccurrence.

Abstract

This paper develops a tool for image texture analysis called a Generalized Cooccurrence Matrix, or GCM. GCM's were first introduced by Davis et al as a general-purpose tool for image texture description. They are based on a model of image texture as a two-dimensional arrangement of texture elements, or primitives. GCM's do not describe textures by directly describing the shape and spatial arrangement of the texture elements. Rather, they describe the spatial arrangement of local image features, such as edges and lines. These local features are generally easier to compute than the texture elements themselves. Section 2 contains a more precise definition of GCM's. Such features are usually physically significant, i.e., they correspond to important physical discontinuities (reflectance, orientation, etc.) in the underlying scene. There is also psychophysical evidence that such local features play a prominent role in human texture perception.

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 1979
Accession Number
ADA084536

Entities

People

  • Larry S. Davis

Organizations

  • University of Texas at Austin

Tags

DTIC Thesaurus Topics

  • Discontinuities
  • Geometry
  • Mathematics
  • Mental Processes
  • Optical Properties
  • Orientation (Direction)
  • Perception
  • Physical Properties
  • Reflectance
  • Two Dimensional

Readers

  • Computer Vision.
  • Ocean-Atmosphere Mesoscale Modeling, Data Assimilation, and Flux Boundary Layers