Multisensor Surveillance Data Augmentation and Prediction

Author

Garret Thayer Bush III, University of Rhode Island

Date of Award

1980

Degree Type

Dissertation

Degree Name

Doctor of Philosophy in Ocean Engineering

Department

Ocean Engineering

First Advisor

Lester LeBlanc

Abstract

In this study the spatial characteristics of an oil spill on the high seas are examined in the interest of determining whether linear-shift-invariant data processing implemented on an optical computer would be a useful tool in analyzing spill behavior. Simulations were performed on a digital computer using data obtained from a 25,000 gallon spill of soy-bean oil in the open ocean.

1. Marked changes occurred in the observed spatial frequencies when the oil spill was encountered. An optical detector may readily be developed to sound an alarm automatically when this happens.
2. The average extent of oil spread between sequential observations was quantified by a simulation of non-holographic optical computation. Because a zero crossover was available in this computation, it may be possible to construct a system to measure automatically the amount of spread.
3. Oil images were subjected to deconvolutional filtering to reveal the force field which acted upon the oil to cause spreading.
4. Some features of spill-size prediction were observed. Calculations based on two sequential photos produced an image which exhibited characteristics of the third photo in that sequence.

Recommended Citation

Bush, Garret Thayer III, "Multisensor Surveillance Data Augmentation and Prediction" (1980). Open Access Dissertations. Paper 1312.
https://digitalcommons.uri.edu/oa_diss/1312