Date of Award

3-14-2014

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Systems Engineering and Management

First Advisor

David R. Jacques, PhD.

Abstract

Small Unmanned Aerial Systems (SUAS) have been utilized by the military, geological researchers, and first responders, to provide information about the environment in real time. Hyperspectral Imagery (HSI) provides high resolution data in the spatial and spectral dimension; all objects, including skin have unique spectral signatures. However, little research has been done to integrate HSI into SUAS due to their cost and form factor. Multispectral Imagery (MSI) has proven capable of dismount detection with several distinct wavelengths. This research proposes a spectral imaging system that can detect dismounts on SUAS. Also, factors that pertain to accurate dismount detection with an SUAS are explored. Dismount skin detection from an aerial platform also has an inherent difficulty compared to ground-based platforms. Computer vision registration, stereo camera calibration, and geolocation from autopilot telemetry are utilized to design a dismount detection platform with the Systems Engineering methodology. An average 5.112% difference in ROC AUC values that compared a line scan spectral imager to the prototype area scan imager was recorded. Results indicated that an SUAS-based Spectral Imagers are capable tools in dismount detection protocols. Deficiencies associated with the test expedient prototype are discussed and recommendations for further improvements are provided.

AFIT Designator

AFIT-ENV-14-M-70

DTIC Accession Number

ADA599322

Download

Share

COinS