STI College- Tacloban

Research on ARCHER (Airborne Real-time Cueing Hyperspectral Enhanced Reconnaissance) Partial Requirement for Operations Research

Submitted by: Piamonte, Joie Ara Mari C. BSCS-III

Submitted to: Sir Michael A. Ellis Instructor

JULY 1, 2010

Page |1

Table of Contents

Abstract ....3 Overview...4 Introduction.. 5 Technical Description.5

Hardware6 Off-line Sensor Data Processing Tools.....6 Real-time Processing Tools7

Algorithms8
Processing.......8

Page |2

Abstract

Over the past five years, the Air Force Research Laboratory and the Space Computer Corporation have supported the Civil Air Patrol in the development of the Airborne Real-time Cueing Hyperspectral Enhanced Reconnaissance (ARCHER) system, which combines a visible/near infrared hyperspectral imaging system, high-resolution panchromatic imaging sensor, a real-time on-board target detection processor and other supporting elements to aid in civilian search-and-rescue missions. The procurement and fielding of the ARCHER system provided useful practical experience in terms of moving hyperspectral remote sensing into an operational context, including areas where further advances are needed. It also provided an airborne collection platform with which to experiment and refine target detection and change detection algorithms to gain greater insight into detection performance. Many of the system and algorithm characteristics and refinements have been reported previously in the literature. This paper attempts to go one step further in providing several of the key lessons-learned from CAP ARCHER experience, with a specific aim to illuminate the areas where further progress is needed and suggest approaches that may address these remaining shortfalls.

Page |3

Overview
ARCHER is a daytime non-invasive technology, which works by analyzing an objects reflected light. It cannot detect objects at night, underwater, under dense cover, underground, under snow or inside buildings. The system uses a special camera facing down through a quartz glass portal in the belly of the aircraft, which is typically flown at a standard mission altitude of 2500 feet (800 meters) and 100 knots (50 meters/second) ground speed. The system software was developed by Space Computer Corporation of Los Angeles, California and the system hardware is supplied by NovaSol Corp. of Honolulu, Hawaii specifically for CAP. The ARCHER system is based on hyperspectral technology research and testing previously undertaken by the United States Naval Research Laboratory (NRL) and Air Force Research Laboratory (AFRL). CAP developed ARCHER in cooperation with the NRL, AFRL and the United States Coast Guard Research & Development Center in the largest interagency project CAP has undertaken in its 63-year history. Since 2003, almost US$5 million authorized under the 2002 Defense Appropriations Act has been spent on development and deployment. As of January 2007, CAP reported completing the initial deployment of 16 aircraft throughout the U.S. and training over 100 operators, but had only used the system on a few search and rescue missions, and had not credited it with being the first to find any wreckage. In searches in Georgia and Maryland during 2007, ARCHER located the aircraft wreckage, but both accidents had no survivors, according to Col. Drew Alexa, director of advanced technology, and the ARCHER program manager at CAP. An ARCHER equipped aircraft from the Utah Wing of the Civil Air Patrol was used in the search for adventurer Steve Fossett in September 2007. ARCHER did not locate Mr. Fossett, but was instrumental in uncovering eight previously uncharted crash sites in the high desert area of Nevada, some decades old. In addition to the primary search and rescue mission, CAP has also tested additional uses for ARCHER. For example, an ARCHER equipped CAP GA8 was used in a pilot project in Missouri in August 2005 to assess the suitability of the system for tracking hazardous material releases into the environment, and one was deployed to track oil spills in the aftermath of Hurricane Rita in Texas during September 2005.

Page |4

ARCHER
(Airborne Real-time Cueing Hyperspectral Enhanced Reconnaissance)

Introduction
The Airborne Real-time Cueing Hyperspectral Enhanced Reconnaissance, also known by the acronym ARCHER, is an aerial imaging system that produces ground images far more detailed than plain sight or ordinary aerial photography can. It is the most sophisticated unclassified hyperspectral imaging system available, according to U.S. Government officials. ARCHER can automatically scan detailed imaging for a given signature of the object being sought (such as a missing aircraft), for abnormalities in the surrounding area, or for changes from previous recorded spectral signatures. SCC developed ARCHER for the Civil Air Patrol (CAP) The US Air Force Auxiliary. The system is deployed in the GA-8 Airvan, an 8-passenger fixed-gear high-wing monoplane manufactured by Gippsland Aeronautics of Australia. ARCHER is a very intuitive graphical user interface (GUI) wrapped around a suite of SCCs HSI data analysis and visualization tools, each customized for the particular demands of CAP missions. Although primarily intended for enhanced search and rescue capabilities, CAP has found many additional uses for ARCHER, including missions for homeland security, disaster assessment, and drug interdiction.

Page |5

Technical Description
The ARCHER system includes a portable ground-based processor (the "ground station," shown below) that allows post-flight data analysis. Post-flight analysis runs at speeds up to 3x real time with no degradation in algorithm performance. The ground station is also used extensively for training, since it has all the processing capabilities of the airborne system and more.

Hardware ARCHER contains an advanced hyperspectral imaging (HSI) system and a panchromatic high-resolution imaging (HRI) camera. At a standard mission altitude of 2500 feet AGL and 100 knot groundspeed, the HIS system resolution is one square meter per pixel. The HRI camera resolution is about 8 cm x 8 cm (3 in x 3 in) per pixel. ARCHER also contains a global positioning system (GPS) and inertial navigation system (INS). Together, these components provide aircraft location, altitude, pitch, yaw, and roll so that each image pixel can be accurately positioned (geo-registered) on a virtual map, in real time, during a mission. Hyperspectral Imager (HSI) The HSI is based on a grating imaging spectrometer design using a 2/3 CCD camera that has a 16 mm focal length lens. The sensor splits the visible and infrared light spectrum into 240 bands with a spectral resolution better than 2.5 nm within the 445-900 nm range. It has a 10-bit per band radiometric resolution, and a swath width of 752 pixels. Inertial Navigation System(INS) An inertial navigation system (INS) is a navigation aid that uses a computer, motion sensors (accelerometers) and rotation sensors (gyroscopes) to continuously calculate via dead reckoning the position, orientation, and velocity (direction and speed of movement) of a moving object without the need for external references.

High-resolution imager(HRI)
A high-resolution imaging (HRI) black-and-white, or panchromatic, camera is mounted adjacent to the HSI camera to enable both cameras to capture the same reflected light. The HRI camera uses a pushbroom

Page |6

approach just like the HSI camera with a similar lens and slit arrangement to limit the incoming light to a thin, wide beam. However, the HRI camera does not have a diffraction grating to disperse the incoming reflected light. Instead, the light is directed to a wider CCD to capture more image data. Because it captures a single line of the ground image per frame, it is called a line scan camera. The HRI CCD is 6,144 pixels wide and one pixel high. It operates at a frame rate of 720 Hz. At ARCHER search speed and altitude (100 knots over the ground at 2,500 ft AGL) each pixel in the black-and-white image represents a 3 inch by 3 inch area of the ground. This high resolution adds the capability to identify some objects.

Off-line Sensor Data Processing Tools

The off-line processing tools run in batch mode under either the Unix/Linux or Windows operating system. All required inputs are provided interactively at program initiation. After the necessary inputs are provided, the tools run unattended until all required output files are produced. Thus, they are especially useful for surveying and processing large amounts of collected data. ---------------------------------------------------------GeoSharpenTM The GeoSharpenTM tool converts hyperspectral and panchromatic data into highresolution, geo-registered survey imagery. It takes as its input any or all of the data files collected during a run and produces a series of geo-registered, pan-sharpened images. These images have the high resolution of the panchromatic data and color taken from the hyperspectral data. Any three hyperspectral bands can be selected for the RGB channels, and the high-resolution pan data are used to sharpen them.

GeoRegTM The GeoRegTM tool accepts HSI or panchromatic data and produces three types of geo-registered outputs: geo-registered imagery using interpolation; geo-registered imagery using a nearest-pixel approach; IGM files for ENVI users. Program input can be any or all of the HSI data files collected during a run. For output the user can select either all the hyperspectral bands (for analysis) or just three bands (for RGB display). Output can be either 2-byte integers or appropriately-scaled bytes. The former is best for hyperspectral analysis, the latter for display. The output consists of a series of geo-registered files and their corresponding ENVI headers for all input files processed, at a user-selectable ground projection resolution. GeoChangeTM The GeoChangeTM tool is a data stream processor that co-registers overlapping HSI datasets and performs pixel-level spectral difference evaluation to detect changes between the datasets. Detected spectral differences are passed to SCC's anomaly detection and signature match
Page |7

detection algorithms. Spectral differences that meet user-defined criteria are declared as targets. GeoChangeTM runs in conjunction with SCC's GeoPaint HSI visualization window, and provides output to SCC's GeoViewTM target evaluation tool.

Real-time Processing Tools Space Computer Corporation's real-time HSI processing tools enable the hyperspectral sensor operator to immediately see HSI sensor data mapped in stable, geo-registered visual form. Targets are detected and declared in real time, and target images that have been sharpened with higher-resolution panchromatic data are typically overlaid on the geo-registered HSI data map. Larger images of targets are displayed to allow immediate visual interpretation and evaluation for relevance to mission objectives. ---------------------------------------------------------GeoPaint SCCs GeoPaint tool is a graphical user interface designed for the very rapid visualization of HSI data. GeoPaint receives HSI data from a hyperspectral sensor concurrent with position and attitude information for the deployment vehicle. The HSI data are geo-registered and mapped, in real time, at their correct locations on a Universal Transverse Mercator grid. GeoPaint offers significant benefits over the sensor coordinate-based presentation that was the only real-time imagery presentation method available before SCC released its GeoPaint tool.
Benefits of GeoPaint Software-based Geo-registration

Attribute Imagery Orientation Imagery stability Imagery Clarity

GeoPaint Real-time Geo-Registered Display Up is always true North. Imagery remains still and stable to allow visual assessment; image shifts in discrete amounts to keep leading edge within viewing window. Imagery appears as it actually exists, without smearing and spatial distortion. Software geo-registration eliminates the need for

Traditional Waterfall Display Up is always the current movement direction of the host vehicle. Imagery is always cascading down the window. The constant movement makes image assessment very difficult because the eye must track the imagery movement. Imagery is smeared and distorted as the sensor undergoes abrupt movements in response to air turbulence. More moving parts means higher maintenance costs, more system failures, and lost deployments.

Fewer closed-loop gimbal position measurement and moving parts pointing.

GeoViewTM SCCs GeoViewTM is a graphical user interface designed for rapid visual evaluation of targets detected in HSI data. GeoViewTM receives data for declared targets from SCCs target detection algorithms. Targets are processed and presented for visual inspection, interpretation, and evaluation in the GeoViewTM window.

Page |8

Target Acquisitions and Identification ARCHER executes three separate algorithms for target acquisition and identification. Spectral signature matching: ARCHER compares reflected electromagnetic radiation (EMR) against a library of spectral signatures to identify specifically targeted objects. Anomaly target detection: ARCHER compares reflected EMR against a continuously calculated background spectrum. Spectral anomalies are flagged as potential targets for further evaluation. Change detection: Using reflected EMR, ARCHER executes a pixel-by-pixel comparison of current ground conditions against ground conditions that were obtained in a previous mission over the same area. Scene changes are identified: new targets, departed targets, and moved targets are highlighted for evaluation.

Real Time Hyperspectral Detection Algorithms In ARCHER, an onboard signal processor accepts blocks of hyperspectral data that are used to estimate radiance statistics of the physical background (usually located beneath an airborne collection platform). These statistics are required for constructing all the onboard real time detection algorithms. Anomaly Detection The RX Anomaly Detector The standard (RX) anomaly detector is a test for rare pixels, as determined by the model pdf of Equation (2). Unusual pixel valuescorresponding to a pdf value less than some threshold valueare labeled targets. This decision is logically equivalent to a test of the form

x C 1 x k threshold
for deciding that a pixel with radiance vector x contains a target. The values of and C are defined in the Equation ; no distinction is made between their estimated and true values. Signature Match Target Detection The signature-match detection algorithm detects pixel spectra in the scene that are similar to one or more given spectra. The algorithm receives HSI data from the sensor. Each incoming pixel spectrum is assigned a score that indicates how closely it matches each of the given spectra. Pixels with scores greater than a threshold are then processed through SCC's spatial-clustering algorithm before declaration as a target. Spectral signatures may be entered as laboratory- or field-grade spectral measurements or they may be harvested from any target that has been detected in an HSI dataset.

Searching for Blue Tarps Using a Spectral Signature

Page |9

Target detection by spectral signature matching can be accomplished by using a laboratory- or field-collected spectral signaure, as shown at left, or by using an in-scene signature harvested from a detected target, as shown at right. Laboratory-grade spectral signature of a typical blue tarp. In-scene spectral signature of a typical blue tarp collected from a detected anomaly target.

Change Detection tries to identify changes in the probability distribution of a stochastic process or time series. In general the problem concerns both detecting whether or not a change has occurred, or whether several changes might have occurred, and identifying the times of any such changes.

Processing A monitor in the cockpit displays detailed images in real time, and the system also logs the image and Global Positioning System data at a rate of 30 Gigabytes (GB) per hour for later analysis. The on-board data processing system performs numerous real-time processing functions including data acquisition and recording, raw data correction, target detection, cueing and chipping, precision image geo-registration, and display and dissemination of image products and target cue information.

P a g e | 10

Conclusion

The Airborne Real-time Cueing Hyperspectral Enhanced Reconnaissance (ARCHER) , is an aerial imaging system that produces highly-detailed ground images. It is has direct applications for search and rescue, counterdrug, disaster relief and impact assessment, and homeland security, and has been deployed by the Civil Air Patrol (CAP) in the United States on the Australian built Gippsland GA8 Airvan fixed-wing aircraft. ARCHER makes use of its hyperspectral imaging capability to automatically scan detailed
imaging for a given signature of the object being sought . Thanks to the support and funding of CAP,

aerial imaging system has been made possible, by which through this search and rescue, counterdrug, disaster relief and impact assessment, and homeland security have been made within reach.

P a g e | 11

Definition of Technical Terms Airborne flying; in the air (Engineering / Aeronautics) (of aircraft)

Degrade Deploy To position (troops) in readiness for combat, as along a front or line; To bring (forces or material) into action. To base (a weapons system) in the field. to reduce in worth, character, etc.; disgrace; dishonour to reduce in rank, status, or degree; remove from office; demote to reduce in strength, quality, intensity, etc.

Hyperspectral Imagery Term used to describe the imagery derived from subdividing the electromagnetic spectrum into very narrow bandwidths. These narrow bandwidths may be combined with or subtracted from each other in various ways to form images useful in precise terrain or target analysis. Also called HSI.

Hyperspectral Imaging System An infrared imaging system that has more than 30 spectral channels with relatively fine spectral resolution, allowing imaging spectroscopy to be carried out.

Envi Image Files ENVI uses a generalized raster data format consisting of a simple flat binary file and a small associated ASCII (text) header file. This approach permits ENVI's flexible use of nearly any image format. All data types are supported in their native formats .

Geographic Information System(GIS) A geographic information system is a system designed to capture, store, manipulate, analyze, manage, and present all types of geographical data.

P a g e | 12

Global Positioning System(Gps) The Global Positioning System (GPS) is a space-based satellite navigation system that provides location and time information in all weather, anywhere on or near the Earth, where there is an unobstructed line of sight to four or more GPS satellites.

Panchromatic made sensitive to all colours by the addition of suitable dyes to the emulsion Compare (Miscellaneous Technologies / Photography) Photog (of an emulsion or film)

Procure To get by special effort; obtain or acquire

Quartz Glass a colourless glass composed of almost pure silica, resistant to very high temperatures and transparent to near-ultraviolet radiation Sometimes shortened to quartz

Raster Data Format Raster data models incorporate the use of a grid-cell data structure where the geographic area is divided into cells identified by row and column. This data structure is commonly called raster. While the term raster implies a regularly spaced grid other tessellated data structures do exist in grid based GIS systems.

Real-Time denoting or relating to a data-processing system in which a computer receives constantly changing data, such as information relating to air-traffic control, travel booking systems, etc., and processes it sufficiently rapidly to be able to control the source of the data (Electronics & Computer Science / Computer Science)

Reconnaissance is the military term for exploring beyond the area occupied by friendly forces to gain information about enemy forces or features of the environment.

Signature a distinctive mark, characteristic, etc., that identifies a person or thing.

P a g e | 13