Date of Award
Master of Science
Department of Electrical and Computer Engineering
Michael L. Talbert, PhD
The focus of this research is to determine if applying edge detection segmentation (proposed by Ramin Zabih, Justin Miller, and Kevin Mai) to Unmanned Aerial Vehicle (UAV) video footage can provide meaningful segments for database storage and retrieval. The edge detection segmentation algorithm is applied to fifty-four UAV video sequences containing visual effects such as abrupt camera changes, camera zooms, motion (rapid and gradual), and cloud cover while varying the frame rate from 5 fps to 30 fps. An analysis of the results is performed to compare actual versus expected outcomes, similar sequences, and scenes with motion, along with explaining false positives/anomalies. Although the frame rate variation and analysis of the scenes with cloud cover are inconclusive, applying the edge detection segmentation algorithm to abrupt changes, rapid motion, and camera zooms produced favorable results, as these were all detected as scene changes. Several near-term and long-term benefits can be drawn from these results, and are provided at the conclusion of the paper, along with recommendations for future research.
DTIC Accession Number
Pyburn, Bradley L., "Analysis of the Applicability of Video Segmentation to Unmanned Aerial Vehicle Surveillance Video" (1999). Theses and Dissertations. 5227.