Automatic Target Recognition

 

Sponsor         Dr. Tom Cooley,    thomas.cooley@kirtland.af.mil

 

Liaison:        Hyperspectral Imaging division

                      Sensors Directorate

                      Hanscom Air Force Base

 

Information: United States Air Force

                      Battlespace Surveillance Innovation Center

                      Hyperspectral Exploitation Program        

 

Team Contact:    Marisol A. Buelow, mab97@dana.ucc.nau.edu

                                               

Project Overview

The Automatic Target Recognition (ATR) project will use image processing algorithms to analyze hyperpecral images to find and classify hostile vehicles.  The first classification process uses the Hough Transform algorithm to locate lines over hostile terrain.  Lines above specified threshold values are then classified as tracks left behind moving enemy vehicles. Areas on and around the tracks are then searched for tanks and trucks using hyperspectral technology to locate and classify hostile targets. 

The project is a cross-disciplinary project between Electrical Engineering and Computer Science; the Electrical Engineering subteam will be working on sensor and systems design during the Fall 2004 semester, with the Computer Scientists coming onboard in January 2005 to design and implement the computational elements.


 

Prerequisites:        Required: Understanding of linear algebra and some knowledge of image processing.

                                Ability to read and code in C++.  Familiarity with Microsoft Windows operating system.

                                Helpful: Familiarity with Matlab and/or IDL languages. 


 

Equipment              Required: The team should have access to a computer that has a C++ compiler and

Requirements:         Matlab programs installed.  Systems should run Windows 2000 or higher. 


     Deliverables:         The student team will be expected to propose and deliver:

                                   *An acceptable solution to the problem that works correctly on a Windows platform

                                    *Demonstrations and regular status reports to sponsor