Date of Award

5-2016

Degree Type

Thesis

Degree Name

Master of Science

Major

Geography

Major Professor

Budhenrda L. Bhaduri

Committee Members

Nicholas Nagle, Bruce Ralston

Abstract

Geospatial data of national infrastructure are a valuable resource for visualization, analysis, and modeling. Building these geospatial foundation-level infrastructure datasets presents numerous challenges. Among those challenges is that of acquiring non-visible attribution of particular infrastructure entities for which there is no viable tabular source. In the case of electric power transmission lines, these data are difficult to acquire, particularly nation-wide. The route, or geometry of transmission lines can be determined from aerial imagery, but nominal voltage, a fundamental requirement for analysis and modeling, is not readily apparent. However, inferences can be made about the nominal voltage based on visual characteristics, or predictors. This study develops a methodology to extract predictors from high-resolution aerial imagery and test the efficacy of those predictors for classifying the nominal voltage of transmission lines using a supervised classifier.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS