Date of Award

3-13-2009

Document Type

Thesis

Degree Name

Master of Science in Electrical Engineering

Department

Department of Electrical and Computer Engineering

First Advisor

Michael J. Mendenhall, PhD

Abstract

Synthetic Aperture Radar (SAR) can generate high resolution imagery of re- mote scenes by combining the phase information of multiple radar pulses along a given path. SAR based Intelligence, Surveillance, and Reconnaissance (ISR) has the advantage over optical ISR that it can provide usable imagery in adverse weather or nighttime conditions. Certain radar frequencies can even result in foliage or limited soil penetration, enabling imagery to be created of objects of interest that would otherwise be hidden from optical surveillance systems. This thesis demonstrates the capability of locating stationary targets of interest based on the locations of their shadows and the characteristics of pixel intensity distributions within the SAR imagery. Shadows, in SAR imagery, represent the absence of a detectable signal reflection due to the physical obstruction of the transmitted radar energy. An object's shadow indicates its true geospatial location. This thesis demonstrates target detection based on shadow location using three types of target vehicles, each located in urban and rural clutter scenes, from the publicly available Moving and Stationary Target Acquisition and Recognition (MSTAR) data set. The proposed distribution characterization method for detecting shadows demonstrates the capability of isolating distinct regions within SAR imagery and using the junctions between shadow and non-shadow regions to locate individual shadow-casting objects. Targets of interest are then located within that collection of objects with an average detection accuracy rate of 93%. The shadow-based target detection algorithm results in a lower false alarm rate compared to previous research conducted with the same data set, with 71% fewer false alarms for the same clutter region. Utilizing the absence of signal, in conjunction with surrounding signal reflections, provides accurate stationary target detection. This capability could greatly assist in track initialization or the location of otherwise obscured targets of interest.

AFIT Designator

AFIT-GE-ENG-09-12

DTIC Accession Number

ADA497637

Download

Share

COinS