detection of obscured targets: signal processing james mcclellan and waymond r. scott, jr. school of...
TRANSCRIPT
Detection of Obscured Targets: Signal Processing
James McClellan and Waymond R. Scott, Jr. School of Electrical and Computer Engineering
Georgia Institute of TechnologyAtlanta, GA 30332-0250
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 2
Outline
Introduction 3-D Quadtree Imaging
GPR processing examples
Location with Acoustic/Seismic Arrays Small Number of Receivers (moving) Cumulative Array strategy for imaging
Accomplishments/Plans
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 3
Three Sensor ExperimentSensor Adjustments and Features
Adjustable Parameters for all three sensors Frequency range Frequency Resolution Spatial Resolution Integration time/bandwidth Height above ground
Location
Possible Features for sensors EMI
Relaxation frequency Relaxation strength Relaxation shape Spatial response
GPR Primary Reflections Multiple Reflections Depth Spatial Response
Seismic Resonance Reflections Dispersion Spatial response
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 4
Multi-Resolution Processing
GPR
EMI
Seismic
Imaging
SigProc
Imaging
Features
Features
Features
DecisionProcess
ExploitCorrelation
Training
Detect
Classify
ID
Quadtree Imaging@ increasingResolution(Eliminate Areas)
Target Localization@ specific sites
Ad-HocArray
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 5
Quadtree BackProjection
Sub-Aperture Formation
(Virtual Sensor)
Image Patch Dividing
(Sub-Patch )
Quadtree BackProjection
1st Stage
2nd Stage
3rd Stage
4th Stage
Space-Time Domain Decomposition Image Patch Dividing and Sub-Aperture Formation (Virtual Sensor)
Divide and Conquer Strategy
Computational Complexity O(N2log2N)
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 6
GPR Processing
Data taken in frequency domain with network analyzer 500 MHz to 8 GHz
Imaging Quadtree
Multi-resolution Approximate Backprojection
2-D, extend to 3-D Fast like -k algorithms
y
x
2 =
4.5
"A
w = 3mm
2 =
62.
4mil
a
L = 6.75"Antenna Shape
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 7
Detection and Region Elimination
Purpose : Distinguish regions consisting coherent scatterers, such as mines, from other regions consisting of only clutter.
Detector exploits the fact that from stage to stage the energy for a coherent target increases relative to total energy while energy of clutter decreases.
)(
)(1
1)|(i
i
S
Sii E
ESSP
Define the ratio of the energy in a sub-volume Si+1 to that in its parent’s Si :
The probability of a target being present in stage i+1 is then given by Bayes Rule:
)()....|().|()( 1111 SPSSPSSPSP iiiii
Initial Simulated Data
Y(c
m)
0 10 20 300
10
20
30
Stage 1 Image
0 10 20 300
10
20
30
Stage 2 Image
0 10 20 300
10
20
30
Stage 3 Image
X(cm)
Y(c
m)
0 10 20 300
10
20
30
Stage 4 Image
X(cm)0 10 20 30
0
10
20
30
X(cm)
Selected Areas
0 10 20 300
10
20
30
0
10
20
30
40
05
1015
2025
30350
5
10
15
20
25
X
Likelihood over the surface--two targets at obj1 = [21;4;-20]; and obj2 = [5;27;-20];
Y
Like
lihoo
d
Then apply a threshold test:
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 8
Quadtree Results
Experimental Data from the model mine field
Target : A VS-50 AP Mine buried 1.3 cm deep
010
2030
40
010
2030
400
5
10
15
20
25
X
Likelihood over the surface--REAL DATA obj = [26,11,-1.3];
YLi
kelih
ood
Experimental Raw Data
Y(c
m)
0 10 20 300
10
20
30
Stage 1 Image
0 10 20 300
10
20
30
Stage 2 Image
0 10 20 300
10
20
30
Stage 3 Image
X(cm)
Y(c
m)
0 10 20 300
10
20
30
Stage 4 Image
X(cm)0 10 20 30
0
10
20
30
X(cm)
Selected Areas
0 10 20 300
10
20
30
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 9
3D Detection Volume
0
5
10
15
0
5
10
15-16
-14
-12
-10
-8
-6
-4
-2
0
X(cm)
3D view of the detection volume
Y(cm)
Dep
th(c
m)
051015
-16
-14
-12
-10
-8
-6
-4
-2
0
X(cm)
Side view of detection volume
Y(cm)
Dep
th(c
m)
0 5 10 15051015
-16
-14
-12
-10
-8
-6
-4
-2
03D view of the detection volume
X(cm)
Dep
th(c
m)
Y(cm)
Synthetic Data: 1 target at position X = 13, Y = 10, Z = -10
Four quadtree stages, 1 cm step size
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 10
Outline
Introduction 3-D Quadtree Imaging
GPR processing examples
Location with Acoustic/Seismic Arrays Small Number of Receivers (moving) Cumulative Array strategy for imaging
Accomplishments/Plans
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 11
Seismic Sensor
R adar:R .F. Source,D em odulator, andS ignal P rocesssing
S ignal G enerator
E lastic W aveTransducer
E lasticSurfaceW ave
M ine
E .M . W aves
A irSoil
S N S
Wav
egui
deD isp lacem ents
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 12
Home in on a Target
Problem Statement: Use a small number of source & receiver positions to locate
targets, i.e., landmines Minimize the number of measurements
Three phases1. Probe phase: use a tiny 2-D array (rectangle or cross)
Find general target area from reflected waves
2. Adaptive placement of additional sensors Maneuver receiver(s) to increase resolution RELAX/CLEAN recalculates for cumulative sensor array
3. On-top of the target Verify the resonance
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 13
Adaptive Sensor Placement
Probe ArrayProbe Array findsgeneral target area
1
2
3Additional sensors are added to the “cumulative array”
Target
Source
On-top forresonance
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 14
Wideband RELAX algorithm
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 15
Surface Plot for Displays
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 16
Examples using Sandbox Data
Single source is used Only the Reverse wave is used in processing
Or, passive listening for Buried structures Remove the Forward wave
Prony analysis, or spatial filter Frequency range
obtained from Spectrum Analysis of measured data Future work:
use RELAX to separate Forward and Reverse waves Cylindrical wave model
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 17
Spectrum Analysis via Prony
TS-50 at 1cm VS1.6 at 5cm
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 18
Processing Examples
Probe Phase Pick five sensors in a cross pattern Apply CLEAN/RELAX algorithm and plot the
“RELAX” surface over a search grid Maneuver Phase
Add one more sensor at a time Increase Aperture, or Triangulate Spatial resolution increases which narrows
down the search area sfor the target
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 19
TS-50 (1cm)
Source at (-20,50)
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 20
Move Direction after PROBE: Three likely directions to move when adding one more sensor
Next Measurement
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 21
1
2
3
Next Measurement
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 22
Further Probing of Direction-1
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 23
Further Probing of Direction-3
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 24
Another Probing of Direction-1
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 25
Sensor is placed at the Peak of Previous Step
On-Top
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 26
Yet Another Strategy
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 27
Accomplishments Developed three sensor experiment to study multimodal processing
Developed new metal detector and a radar Investigated three burial scenarios Showed responses for all the sensors over a variety of targets Demonstrated possible feature for multimodal/cooperative processing Developed new 3D quadtree strategy for GPR data
Developed seismic experiments, models, and processing Improved experimental measurement by incorporating a Wiener filter Demonstrated reverse-time focusing and corresponding enhancement of mine signature Demonstrated imaging on numerical and experimental data from a clean and a cluttered environment Modified time-reverse imaging algorithms to include near field DOA and range estimates. The algorithms are
verified for both numerical and experimental data with and without clutter. Modified wideband RELAX and CLEAN algorithms for the case of passive buried targets. The algorithms are
verified for both numerical and experimental data with and without clutter. Used RELAX imaging to locate targets with cumulative maneuvering receivers. Developed a vector signal modeling algorithm based on IQML (Iterative Quadratic maximum Likelihood) to
estimate the two-dimensional -k spectrum for multi-channel seismic data. Developed multi-static radar
Demonstrated radar operation with and without clutter objects for four scenarios Buried structures
Developed numerical model for a buried structure Demonstrated two possible configurations for a sensor Made measurement using multi-static radar
MURI Review 13-Jan-05 Scott/McClellan, Georgia Tech 28
Plans
Three sensor experiment (Landmine) Incorporate reverse-time focusing and imaging Incorporate multi-static radar More burial scenarios based on inputs from the signal processors
More/Stronger clutter Distribution of targets and clutter Close proximity between clutter and targets
Look for more connections between the sensor responses that can be exploited for multimodal/cooperative imaging/inversion/detection algorithms
Imaging/inversion/detection algorithms Extend 3-D quadtree algorithm to multi-static GPR data. Investigate the use of reverse-time ideas to characterize the inhomogeneity of the ground Investigate the time reverse imaging algorithm for multi-static GPR data. Investigate the CLEAN and RELAX algorithms for target imaging from reflected data in the
presence of forward waves with limited number of receivers. Investigate joint imaging algorithms for GPR and seismic data.
Buried Structures Experiments with multi-static radar Develop joint seismic/radar experiment Signal Processing