Date of Award

1984

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Abstract

The problem of texture measurement definition is studied. A theory is presented in order to overcome the problem. This theory includes a conceptual framework for the general measurement definition problem. This framework is based on the concepts of a perceptual transform, a perceptual space, a measurement transform and a similarity transform. Using this conceptual framework and a particular choice for the similarity transform, a set of requirements are defined. These requirements can be used to create a formal method for defining measurements off the GLC matrices. The formal procedure is based on preserving perceptual relationships among textures. To apply this technique a perceptual norm, a least-squares procedure and a synthesis procedure are needed. Each of these components are investigated and the motivation for selecting each of the methods used is presented. Individually, these components are tested in order to see if they are appropriate for use with our technique. Finally, a feasibility study is discussed to demonstrate the possibility of using this technique to solve for measurements. The problem was to define measurements given a limited number of textures. The defined measurements are studied in order to establish their contributions. It is shown that the newly defined measurements are gauging the periodicity and the symmetry of patterns, perceptual entities that have been demonstrated to be important in the human vision system.

Pages

394

Share

COinS