Trevor Clarke

Abstract

Wide area airborne surveillance (WAAS) systems are a new class of remote sensing imagers which have many military and civilian applications. These systems are characterized by long loiter times (extended imaging time over fixed target areas) and large footprint target areas. These characteristics complicate moving object detection and tracking due to the large image size and high number of moving objects. This thesis evaluates existing object detection and tracking algorithms with WAAS data and provides enhancements to the processing chain which decrease processing time and increase tracking accuracy. Decreases in processing time are needed to perform real-time or near real-time tracking either on the WAAS sensor platform or in ground station processing centers. Increased tracking accuracy benefits real-time users and forensic (off-line) users. The original contribution of this thesis increases tracking efficiency and accuracy by breaking a WAAS scene into hierarchical areas of interest (AOIs) and through the use of hyperspectral cueing.

Library of Congress Subject Headings

Remote sensing--Data processing; Computer vision; Multispectral photography--Data processing

Publication Date

2010

Document Type

Thesis

Department, Program, or Center

Computer Science (GCCIS)

Advisor

Canosa, Roxanne

Advisor/Committee Member

Gaborski, Roger

Advisor/Committee Member

Bailey, Reynold

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TA1637 .C53 2010

Campus

RIT – Main Campus

Download

Share

COinS