EECS 826 – InSAR and Applications Utilizing SAR Systems for Landmine Detection EECS 826 Jason Kieffaber April 28, 2009.

EECS 826 – InSAR and Applications

Utilizing SAR Systems for Landmine Detection

EECS 826Jason Kieffaber

April 28, 2009

EECS826 – SAR Systems for Landmine Detection

Overview

Agenda• Overview• GPR and UWB Basics

• GPR Basics• UWB Basics• UWB GPR Systems

• Current Systems • Rail-GPSAR• BoomSAR• REMIDS

• Summary• Questions and References


Overview

Background (Costs)• 110 Million deployed

• 60+ countries• 100+ Million awaiting deployment

• Deployment cost: $3-$30• Clearing cost: $300-$1000• 2000 people/month (mostly civilians)• Survivor costs: $5000• Viable for at least 50 years• 1100 years to clear


Overview

Background (Land Mines)• Anti-Tank Landmines

• Buried• Larger/Uniform• War-time Use

• Anti-Personnel Landmines• Surface Placement• Foliage Cover• Smaller/Less Uniform• Civilian Use/Targets


Overview

Background (Clearing)• War-Time

• Military Focus• Anti-Tank Emphasis• Forward Looking• Speed

• Post-War• Humanitarian Focus• Anti-Personnel • Longer Time-Scale

• Removal vs. Intentional Detonation


GPR and UWB Basics

Ground Penetrating and Ultra Wide-Band RADAR• GPR Basics

• Acquisition Platform• Propagation• Target Recognition

• UWB Basics• Frequency and Bandwidth

• GPR UWB Systems Summary


GPR and UWB Basics

Acquisition Platform• Common GPR Methodology

• Reflection Seismology• Antenna/Ground Coupled• Time Consuming• Dangerous for Mines


GPR and UWB Basics

Acquisition Platform (Ground Based)• Near-Earth

• Metal Detector Coupled• Vehicle Mounted

• Forward Looking• Boom Mounted

• Negatives• Still Dangerous• Incomplete Coverage• Real-Time Requirements• No Ground Contact• Time Consuming (False Positives)


GPR and UWB Basics

Acquisition Platform (Airborne)• Common SAR Methods

• Safe (Generally)• Real-Time Not Required• Faster/More Coverage• Accessibility• Post Analyze False Targets

• Negatives• Fewer Samples• No Ground Contact


GPR and UWB Basics

Propagation• Multiple Reflection Boundaries• Homogeneous Assumption

• Air/Ground Boundary• Ground/Target Boundary• Uniform Soil Type

• Reality• Air/Foliage Boundary• Air/Ground Boundary• Ground/Ground Boundary• Ground/Target Boundary• Generally Non-Uniform Soil


GPR and UWB Basics

Propagation (Soil Dielectrics)• Index of Refraction Tied to Dielectric Constant

• Permeability: µr ≈ 1• Permittivity/Dielectric Constant εr

• Dielectric Constant Depends on Moisture

• Snell’s Law

• Moisture Content Depends on Soil Type• Permeability ≈ 1 (Most Soil)• Complex Relationship• Could be Multiple Boundaries


GPR and UWB Basics

Propagation (Solutions)• Traditional GPR

• Extensive Operator Training• Small Areas, Near Constant Soil Characteristics

• Non-Traditional Options • Remove/Compensate for Boundary Changes

• Air/Ground Relatively Easy, Others Complex• Usually Post-Processing

• Simple Migration Compensation• Complex Dielectric Modeling• Algorithmic Training


GPR and UWB Basics

Propagation (Migration Compensation) [McClellan et. al.]

• Conventional Focusing Techniques Inadequate• Multipath Issues• Non-Hyperbolic Response

• Post Filter Differently• 2-D FIR Filter• Simple Model• Successive Focus Steps


GPR and UWB Basics

Propagation (Modeling) [Jin, Zhou, et. al.]

• ψt = Real{θ t}• Incident Electric Field at Z

• Transmission Factors

• Geometry

COMPLEX!


GPR and UWB Basics

Propagation (Training)• Training

• Apply a Suite of Algorithmic Processes• Requires Post Processing• Requires Multiple/Varying Passes

• Multiple Simple Migration Passes• Index of Complex Soil Dielectrics

• Time and Processing Intensive• Not Applicable for Real Time Applications• Still More Art Than Science


GPR and UWB Basics

Target Recognition• False Positives

• Single Biggest Problem• Rock vs. Junk vs. Bush vs. Mine• Conservative Estimate of 10:1• Wasted Time

• Target Recognition/ROI Identification Improvements• Interference Suppression• Symmetry Recognition• ESD Recognition• Simple Training Methodology


GPR and UWB Basics

Target Recognition (Interference) [Zhang, Song, et. al.]

• Radio Frequency Interference (RFI)• With Small Sample Size, Cannot Accurately Estimate• No Estimation, No Compensation

• Enlarge Sample Size• 2-D Fourier (If Band-limited)• Fan Shaped Regions of Validity• Provides Superior Rejection

• If Band-limited, Can Achieve Better RFI Rejection


GPR and UWB Basics

Target Recognition (Symmetry) [Carin, Ressler, et. al.]

• Assumptions• Device is Symmetrical• Body of Revolution• Primary Axis Perpendicular to Surface• Assumes Uniform Device Conductivity

• Multiple 3-D Algorithms Available• Classic Model – Physical Optics • Method of Moments

• Effective for Buried and Surface Placement


GPR and UWB Basics

Target Recognition (Physical Optics)• Apply Incident Field/Ray Instead of Total Field

• Apply at Multiple Sample Points• Accurate for Small Scatter Relative to Field

• Accurate for Ideal Surfaces• Simple Model• Born Approximation


GPR and UWB Basics

Target Recognition (Method of Moments)• More Accurate Estimation/Modeling

• Also Sample/Phenomenological Based• Polarized Wave-front Analysis

• A) Diffraction (Front)• B) Diffraction (Back)• C) Reverberation• D) Creeping-Wave Circumnavigation• E) Diffraction (Bottom)• F) Diffraction (Multiple)


GPR and UWB Basics

Target Recognition (ESD/Spectral) [Ho, Carin, et. al.]

• Analysis of the Spectral Properties of the Mine• Similar to Propagation Analysis • Works for Metal and Plastic Mines

• Provided Sufficient Dielectric Contrast• Resistant to Differences in Soil Dielectrics• Utilizes 3-D FDTD (Finite-Difference Time Domain)

• Baseline• Builds Database from Measured Models

• d (data) = g (ground bounce) + s (landmine scatter) + w (noise)

• Same Drawbacks as Propagation Analysis


GPR and UWB Basics

Target Recognition (ESD/Spectral)


GPR and UWB Basics

Target Recognition (Training and Detection)• Example Process:

• SVM: Support Vector Machine (Machine Learning)• PCA: Principle Component Analysis

• Effect to be Matched (ESD, Dielectrics, Etc.)


GPR and UWB Basics

Frequency and Bandwidth (Needs)• Surface and Buried Mines

• Foliage AND Ground Penetration• Resilient to Clutter• Band-limited• Frequency Tradeoff

• Higher Frequency = Better Resolution• Lower Frequency = Deeper Ground Penetration

Depth of Penetration

Optimal Frequency

Applications

0-1.5 ft (0-0.5 m)

1600 MHzStructural Concrete,

Roadways, Bridge Decks

0-3 ft (0-1 m)

900 MHz Concrete, Shallow Soils, Archaeology

0-12 ft (0-9 m)

400 MHzShallow Geology, Utilities,

UST's, Archaeology


GPR and UWB Basics

Frequency and Bandwidth (Ultra Wide-Band)• Ultra Wide-Band Definition

• Bandwidth Must be 25% of the Center Frequency

• Band-Limited• Resistant to Clutter Interference• Resistant to Radio Interference• Relatively Low Power• UWB is a Solid Solution


GPR and UWB Basics

Frequency and Bandwidth (Specifics)• GPR Desirable Specifics

• Approximately 1 GHz Center Frequency• Range From 0.1 GHz to 3 GHz• Penetration Depth of Approximately 1 Meter• Effective for Most Mine Types

• Approximately 1 GHz Bandwidth• Range From 0.5 GHz to 2 GHz• 100% BW (UWB)• Effective for Power, SNR, and Clutter Rejection

• Short Pulse (Nanoseconds) Duration


GPR and UWB Basics

GPR UWB Systems Summary• Platform

•Normally Airborne, UWB, Short Pulse• System Training

• Dielectric Soil, Spectral, Discrimination Factors• Image Acquisition

• SAR, Desired System Parameters• Prescreening Stage

• Ground/Air Boundary Rejection, Soil Assumptions• Target Discrimination

• Migration, Clutter Rejection, Target Matching


Current Systems

Current Systems• Rail-GPSAR• BoomSAR• REMIDS• Other Areas of Development


Current Systems

Rail-GPSAR


Current Systems

Rail-GPSAR (Description)• Experimental Landmine Detection System• Rail/Trolley Mounted (24m long, 3.4m high)• 1 TX, 1 RX TEM Horn Antenna (VV/HH Polarization)• Geodimeter Calibrated• Strip-Map, Side-Looking SAR (200 m2)• Image Formation (Constant Integral Angle Back)• Short Pulse UWB • 10 GHz Sampling• Forward Speed of 5cm/s• Center Frequency: 0.3 GHz – 1.9 GHz• Average Power 50 mW


Current Systems

Rail-GPSAR (Block Diagram)


Current Systems

Rail-GPSAR (RFI Suppression)


Current Systems

Rail-GPSAR

(PCA/SVM)


Current Systems

BoomSAR (Description)• Research on FOPEN and GPEN (Phenomenology)• Army Research Laboratory – SAR on a Boom• Built in 1995 – Aberdeen and Yuma Proving Grounds• JLG Boom (1000 pounds, 45m maximum)• Low Speed – 1 km/hr• Geodimeter Calibrated• 2 TX, 2 RX TEM Horn Antennas (VV, HH, VH, HV)• Center Frequency 0.04 GHz - 2 GHz • 2 GHz Sampling• Average Power 1.5 W• UWB Short Pulse• 300 m Range Swath Width• 128 Burst/Second/Polarization


Current Systems

BoomSAR (Block Diagram)


Current Systems

BoomSAR (Example Image)


Current Systems

REMIDS


Current Systems

REMIDS (Description)• Remote Minefield Detection System• Endfire/Forward Looking UWB SAR and Infrared Camera• UK Ministry of Defense (2002)• 80 m Test Track/5 Soil Types• 1 TX, 2 RX Endfire (not Broadside) Antennas

• Left/Right Ambiguity (As Tested)• Degraded Resolution• Aperture Vertically Limited (Not Horizontally)

• 3 GHz Bandwidth (1.5 GHz Center) – Short Pulse• 10 GHz Sampling• Offline Analysis


Current Systems

REMIDS (System Overview)


Overview

REMIDS (Test Track – Thales/Paris)


Overview

REMIDS (Sample Image)


Current Systems

REMIDS (System Diagram)


Current Systems

Other Areas of Development [Optional]• Submerged/Underwater Mines

• Significance• SAR Research/MISE Program

Ultra Wideband Radar Images of the Surface Disturbance Produced by a Submerged, Mine-Like Object. [Sletten]

• Moving Targets • Foliage Penetration• Change Detection

Single-Channel UWB SAR Ground Moving Targets Detection Method Using Change Detection BasedOn Single-Pass Sub-Aperture Images [Zhou, et. al.]


Summary

Agenda:• History and Background• GPR and UWB Basics

• GPR Basics• UWB Basics• UWB GPR Systems

• Current Systems • Rail-GPSAR• BoomSAR• REMIDS


Questions

QUESTIONS?


References• [1] Zhou, Hong, et. al. “Single Channel UWB SAR Ground Moving Targets Detection Method Using Change Detection Based on Single-Pass Sub-Aperture Images” Synthetic Aperture Radar, 2007, APSAR (2007): 266-270.

• [2] Sletten, M.A. “Ultra Wideband Radar Images of the Surface Disturbance Produced by a Submerged, Mine-Like Object.” Geoscience and Remote Sensing 38.6 (2000): 2506-2514.

• [3] Cooper, P. et. al. “Ultra Wideband Endfire Synthetic Aperture Radar for Landmine Detection.” Geoscience and Remote Sensing Symposium (2003).

• [4] Tian, Jin, et. al. “Ultrawideband Synthethic Aperture Radar Landmine Detection.” Geoscience and Remote Sensing 45.11.1(2007): 3561-3573.

• [5] Chant, I.J., et.al. “Overview of Current Radar Land Mine Detection Research at the Defence Science and Technology Organisation, Salisbury, South Austrailia.” Detection fo Abandonded Land Mines: EUREL Conf (Publ. No 431) (1996): 138-142.

• [6] Ressler, M. A. “The Army Research Laboratory Ultra Wideband BoomSAR.” Geoscience and Remote Sensing Symposium 3 (1996): 1886-1888.

• [7] Tian, Jin, et. al. “Ultra-wide Band SAR Subsurface Metallic Landmine Images: Simulation and Measurement.” Intl. Conf. on Radar (2006): 1-4.

• [8] Zhang, Han-hua, et. al. “A Rail-Mounted UWB SAR Landmines Detection System: Rail-GPSAR.” Intl. Conf. on Radar (2006): 5-9.


References• [9] Ho, K.C., et. al. “An Investigation of Using the Spectral Characteristics From Ground Penetrating Radar for Landmine/Clutter Discrimination” Geoscience and Remote Sensing 46.4 (2008): 1177-1191.

• [10] Ressler, M.A., et. al. “Ultra Wideband Radar Images of the Surface Disturbance Produced by a Submerged, Mine-Like Object.” Geoscience and Remote Sensing 35.3 (1997): 762-772.

• [11] McClellan, J.H., et. al. “Migration of Underground Targets in UWB-SAR Systems.” Intl. Conf. Image Processing 1 (2000): 713-716.

• Wikipedia. “Ultra Wideband”, “SAR”, “Endfire”, “BoomSAR”, “Ground Penetrating”, et.al. www.wikipedia.org

• Taylor, James D., “Introduction to Ultra-Wideband Radar Systems”. 1995 • Images:

•http://newsgrist.typepad.com/underbelly/images/landmine2005sm_1.jpg•http://www.state.gov/img/09/30573/090123_fs_demolition_300_1.jpg•http://en.wikipedia.org/wiki/File:Mine_(AWM_304925).jpg•http://www.geocities.com/nhfortress/Towers/mine.jpg•http://upload.wikimedia.org/wikipedia/commons/0/0c/TC24_Italian_landmine_cutaway.png•http://www1.american.edu/TED/ice/images4/ek_landmine.jpg•http://library.thinkquest.org/C008616/site/pics/Landmine.jpg•http://www.betterworldtg.com/en/Stepping_On_Landmine__291w.jpg•http://www.it.uu.se/research/syscon/signalprocessing/topics/NQR/Landmine.jpg

EECS 826 – InSAR and Applications Utilizing SAR Systems for Landmine Detection EECS 826 Jason Kieffaber April 28, 2009.

Documents

eecs826 sar systems

landmine detection gpr

landmine detection eecs

mines slide

uwb basics ground penetrating

clear slide

ground contact slide

nonuniform soil slide