Multispectral Texture.

Abstract

Textures in single-band images are often characterized by statistics of the joint distributions of pairs of gray levels for pairs of pixels in given relative positions, or by statistics of absolute gray level differences for such pairs of pixels. Joint distributions of pairs of spectral vectors in multiband images are cumbersome, since for k bands they are 2k-dimensional; but absolute difference distributions are less so -e.g., for two bands they are only two-dimensional. This paper discusses the possibility of using statistics of absolute difference distributions for characterizing textures in multiband images, with emphasis on the two-band case. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1980
Accession Number
ADA096409

Entities

People

  • Angela Y. Wu
  • Azriel Rosenfeld
  • Cheng-ye Wang

Organizations

  • University of Maryland

Tags

DTIC Thesaurus Topics

  • B Band
  • Computer Science
  • Computer Vision
  • Computers
  • Data Science
  • Histograms
  • Image Processing
  • Inertia
  • Information Science
  • K Band
  • Moment Of Inertia
  • Multispectral
  • Pattern Recognition
  • Statistics
  • Theoretical Computer Science
  • Two Dimensional
  • Universities

Fields of Study

  • Physics

Readers

  • Computer Vision.
  • Statistical inference.