Texture-Tone Study with Application to Digitized Imagery.

Abstract

The report describes initial research efforts undertaken to determine if texture - a significant but ill-defined property of virtually all substances can be used in a practicable automated image interpretation technique. The authors attempt to establish a basis for measures which define the important qualities of texture by first considering discrete tonal features in an image as constituents of areas of homogeneity or lack of homogeneity ('unhomogeneity'). The concept of homogeneity is intended here to signify the existence of groups of similar neighboring resolution cells, which have been judged similar by some prespecified criterion. Homogeneity displays or plots are constructed by linking or connecting the similar neighboring resolution cells and ignoring all others. Unhomogeneity infers a lack of similarity of neighboring resolution cells; and an unhomogeneity plot then, is a display of linked, dissimilar neighboring resolution cells. Digital computer algorithms were written to determine the extent of homogeneity for each resolution cell in a picture, and a group of ten (5 homogeneity and 5 unhomogeneity) on-line plots were made for each of the 36 images. (Author)

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1970
Accession Number
AD0724117

Entities

People

  • Joe D. Bissell
  • Robert M. Haralick

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Computers
  • Digital Computers
  • Homogeneity

Readers

  • Human-Computer Interaction (HCI).
  • Image Processing and Computer Vision.
  • Molecular and Cellular Biochemistry

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference