1. Detection of Obscured Targets: Signal Processing James McClellan and Waymond R. Scott, Jr. School of Electrical and Computer Engineering Georgia Institute of Technology Atlanta, GA 30332-0250 jim.mcclellan@ece.gatech.edu 404-894-8325

2. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech2 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

3. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech3 Three Sensor Experiment Sensor 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

4. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech4 Multi-Resolution Processing GPR EMI Seismic Imaging SigProc Imaging Features Decision Process Exploit Correlation Training Detect Classify ID Quadtree Imaging @ increasing Resolution (Eliminate Areas) Target Localization @ specific sites Ad-Hoc Array

5. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech5 Quadtree BackProjection Sub-Aperture Formation (Virtual Sensor) Image Patch Dividing (Sub-Patch )  Space-Time Domain Decomposition  Image Patch Dividing and Sub-Aperture Formation (Virtual Sensor)  Divide and Conquer Strategy  Computational Complexity O(N 2 log 2 N)

6. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech6 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 Antenna Shape

7. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech7 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. Define the ratio of the energy in a sub- volume S i+1 to that in its parent’s S i : The probability of a target being present in stage i+1 is then given by Bayes Rule: Then apply a threshold test:

8. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech8 Quadtree Results Experimental Data from the model mine field Target : A VS-50 AP Mine buried 1.3 cm deep

9. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech9 3D Detection Volume Synthetic Data: 1 target at position X = 13, Y = 10, Z = -10 Four quadtree stages, 1 cm step size

10. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech10 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

11. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech11 Seismic Sensor

12. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech12 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 phases 1.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

13. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech13 Adaptive Sensor Placement Probe Array Probe Array finds general target area 1 2 3Additional sensors are added to the “cumulative array” Target Source On-top for resonance

14. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech14 Wideband RELAX algorithm

15. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech15 Surface Plot for Displays

16. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech16 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

17. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech17 Spectrum Analysis via Prony TS-50 at 1cmVS1.6 at 5cm

18. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech18 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

19. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech19 TS-50 (1cm) Source at (-20,50)

20. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech20 Move Direction after PROBE: Three likely directions to move when adding one more sensor Next Measurement

21. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech21 1 2 3 Next Measurement

22. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech22 Further Probing of Direction-1

23. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech23 Further Probing of Direction-3

24. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech24 Another Probing of Direction-1

25. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech25 Sensor is placed at the Peak of Previous Step On-Top

26. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech26 Yet Another Strategy

27. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech27 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

28. MURI Review 13-Jan-05Scott/McClellan, Georgia Tech28 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