1 Synthetic Aperture Sonar (SAS) and Acoustic Templates for the Detection and Classification of Underwater Munitions Steven G. Kargl, Kevin L. Williams, Aubrey L. España Applied Physics Laboratory University of Washington 1013 NE 40 th St., Seattle, WA 98105 Raymond Lim, Jermaine L. Kennedy, Rudy T. Arrieta, Timothy M. Marston, Joseph L. Lopes Naval Surface Warfare Center Panama City Division Panama City, FL 32407-7001 Research supported by SERDP and ONR 1
Report Documentation Page Form Approved OMB No. 0704-0188 Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. 1. REPORT DATE NOV 2010 2. REPORT TYPE 3. DATES COVERED 00-00-2010 to 00-00-2010 4. TITLE AND SUBTITLE Synthetic Aperture Sonar (SAS) and Acoustic Templates for the Detection and Classification of Underwater Munitions 5a. CONTRACT NUMBER 5b. GRANT NUMBER 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) 5d. PROJECT NUMBER 5e. TASK NUMBER 5f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) University of Washington,Applied Physics Laboratory,1013 NE 40th Street,Seattle,WA,98105 8. PERFORMING ORGANIZATION REPORT NUMBER 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR S ACRONYM(S) 12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release; distribution unlimited 11. SPONSOR/MONITOR S REPORT NUMBER(S) 13. SUPPLEMENTARY NOTES Presented at the 15th Annual Partners in Environmental Technology Technical Symposium & Workshop, 30 Nov? 2 Dec 2010, Washington, DC. Sponsored by SERDP and ESTCP. 14. ABSTRACT This presentation will present a series of monostatic and bistatic acoustic scattering measurements were conducted to investigate discrimination and classification capabilities based on the acoustic response of targets for underwater unexploded ordnance (UXO) applications. The measurements were performed during March 2010 and are referred to as the Pond Experiment 2010 (PondEx10), where the fresh water pond contained a sand sediment. The measurements utilized a rail system with a mobile tower and a stationary sonar tower. Each tower is instrumented with receivers while the sources are located on the mobile tower. For PondEx10, eleven targets were deployed at two distinct ground ranges from the mobile tower system. Acoustic data were initially processed using synthetic aperture sonar (SAS) techniques and the data were further processed to generate acoustic templates for the target strength as a function of frequency and aspect angle. Preliminary results of the processing of data collected from proud targets are presented. Also presented are the results associated with a processing technique that permits isolation of the response of an individual target, which is in close proximity to other targets. [Research supported by The Strategic Environmental Research and Development Program under projects MR-1665 and MR-1666 and the Office of Naval Research.] 15. SUBJECT TERMS 16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF ABSTRACT a. REPORT unclassified b. ABSTRACT unclassified c. THIS PAGE unclassified Same as Report (SAR) 18. NUMBER OF PAGES 29 19a. NAME OF RESPONSIBLE PERSON
Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std Z39-18
Military Munitions in the Underwater Environment Technical Session No. 4A SYNTHETIC APERTURE SONAR (SAS) AND ACOUSTIC TEMPLATES FOR THE T DETECTION AND CLASSIFICATION OF UNDERWATER MUNITIONS DR. STEVEN KARGL University of Washington Applied Physics Laboratory 1013 NE 40th Street Seattle, WA 98105 (206) 685-4677 kargl@troutmask.apl.washington.edu CO-PERFORMERS: Kevin Williams (University of Washington); Raymond Lim, Jermaine L. Kennedy, Timothy M. Marston, and Joseph L. Lopes (Naval Surface Warfare Center) his presentation will present a series of monostatic and bistatic acoustic scattering measurements were conducted to investigate discrimination and classification capabilities based on the acoustic response of targets for underwater unexploded ordnance (UXO) applications. The measurements were performed during March 2010 and are referred to as the Pond Experiment 2010 (PondEx10), where the fresh water pond contained a sand sediment. The measurements utilized a rail system with a mobile tower and a stationary sonar tower. Each tower is instrumented with receivers while the sources are located on the mobile tower. For PondEx10, eleven targets were deployed at two distinct ground ranges from the mobile tower system. Acoustic data were initially processed using synthetic aperture sonar (SAS) techniques, and the data were further processed to generate acoustic templates for the target strength as a function of frequency and aspect angle. Preliminary results of the processing of data collected from proud targets are presented. Also presented are the results associated with a processing technique that permits isolation of the response of an individual target, which is in close proximity to other targets. [Research supported by The Strategic Environmental Research and Development Program under projects MR-1665 and MR-1666 and the Office of Naval Research.] C-80
2 Outline of Presentation Hypothesis and technical objectives of the research Pond Experiment 2010 (PondEx10) environment Equipment and experimental layout Deployed munitions and scientific targets Experimental results and some data-model comparisons Conclusions and future work 2
3 Hypothesis and Technical Objectives Central hypothesis: The environment can alter the acoustic response of an unexploded ordnance (UXO) significantly, so the environment must be taken into account when developing robust detection and classification strategies. Technical objectives include Understanding the interaction of an acoustic field with a UXO near a watersediment interface over large frequency and aspect angle ranges. Developing an inventory of acoustic responses for various UXOs, which can be used to validate propagation and scattering models. Collecting both monostatic and bistatic synthetic aperture sonar (SAS) data from several targets. These data can be used in standard SAS imaging techniques, in producing acoustic templates (i.e., acoustic finger prints ), in developing and testing of classification algorithms. 3
4 PondEx10 Environment Naval Surface Warfare Center, Panama City Division Test Pond Facility 9 million gallons of fresh water 112 m long, 80 m wide, 14 m deep 1.5 m of sand sediment in the center Filtered and chlorinated water gives 12 m visibility. 4
5 PondEx10 Environment 30.3m 24.5m 24.5m 30.3m Sonar Tower Target Field Patches 25m 14.3m Flat Bottom Area of Test Pool STMS 2 14.3m 25m W S E N NSWC PCD Test Pond Facility Bldg. #383 Panama City, FL 5
6 Equipment and Experimental Layout STMS 2 Rail System NSWC PCD Rail System 21-m rail with 19-m SAS aperture Tower travels at 5 cm/s Sources and receiving array tilted at 20 or 40 depression angle Source waveforms: 1 31 khz LFM Chirp 30 50 khz LFM Chirp 6-channel receiving array Sonar tower positioned at 5 and 10 m ranges during PondEx10 Tilt and pan capability for aligning acoustic axis of receivers to a target Receivers: 1 3 composite transducer ITC 1001 hydrophone 6
7 Equipment and Experimental Layout NSWC PCD Rail System PondEx09 Target Field 10 m Alignment Frame Target A PondEx10 Target Field Guide line at 11 m range 1 2 3 4 5 B 12 m long Direction of Sonar Beam Direction of Sonar Beam 21 m long STMS 2 Rail System 10 m NSWC PCD Sonar Tower 3 m Smoothed Areas 10 m Direction of Sonar Beam 21 m long STMS 2 Rail System One target in the target field One tower scans while other remains stationary Targets rotations: 0 80 and 0 280 with a 20 increment (depends on symmetry of target) Multiple targets in the target field Adjacent targets separated by 1.5 or 3 m Colored patches: 1 1 m 2 smoothed areas Targets rotations: 0 80, -80 80, and 0 280 with a 20 increment (depends on symmetry of target) 7
8 Munitions and Scientific Targets 155 mm howitzer projectile (empty, yellow UXO) 81 mm mortar (filled with cement) 152 mm TP-T round (blue UXO below) Small aluminum cylinder with a notch Solid steel artillery shell Machined aluminum replica of artillery shell Machined steel replica of artillery shell Aluminum pipe (1 ft dia., 2:1 aspect ratio) Solid aluminum cylinder (1 ft dia., 2:1 aspect ratio) Two rocks (i.e., clutter) Alignment frame from PondEx09 and cords registered to frame and STMS 2 rail system 8
9 Five Broadside Targets in PondEx10 Field Machined aluminum replica (top) Solid aluminum cylinder Machined steel replica Aluminum pipe Real steel artillery shell (bottom) Pulse compressed, Baseband Data (a) SAS Images Targets at 10 m range from STMS 2 Rail system (b) Note interference of the scattered time signals from the targets. 0 to -25 db color scale 1. Solid aluminum cylinder in (b) exhibits triplet structure that was recently explained by Williams et al, J. Acoust. Soc. Am., 127, 3356-3371 (2010). 2. Pipe in (d) has a strong echo from acoustic energy transmitted through the pipe and reflected from its back wall. 3. Steel UXO in (e) and steel replica in (c) are similar images while an image for an aluminum replica in (a) is dissimilar. 18.75 m Cross Range 1.2 m (c) (d) (e) 3 ms 2 m 9
10 Five Targets in PondEx10 Field: SAS Images 0 Rotation 20 Rotation 40 Rotation 60 Rotation 80 Rotation Al Replica Al Cylinder Steel Replica Al Pipe 1.2 m 2 m Images normalized to max pixel value in bottom left image: 0 to 25 db range. 10
11 Pulse Compressed Scattered Signals from 5 Targets 0 Rotation 20 Rotation 40 Rotation 60 Rotation 80 Rotation Al Replica of Shell Solid Al Cylinder Steel Replica Aluminum Pipe 18.75 m Cross Range Artillery Shell 3 ms Images normalized to max pixel value in left image: 0 to 40 db range. 11
12 SAS Filtering Algorithm to Resolve Adjacent Targets Cross Range Pulse compressed data Deconvolution with a locus of closest approach arc Form a SAS image of filtered data Cross Range Time axis Filtered Pulse compressed data Target Arc Convolution with locus of closest approach arc Window Image Time axis The deconvolution/convolution processes are linear transformations. The processing does not alter the information in the isolated signals unless the targets are sufficiently close where multiple scattering may become important. 12
13 Acoustic Templates after SAS Filtering 0 Aluminum Replica Steel Replica Steel Artillery Shell 0 90-6.3 Aspect Angle (deg.) 180-12.5 db (Relative) 270-18.3 360 0 8 16 24 32 f (khz) 0 8 16 24 32 f (khz) 0 8 16 24 32 f (khz) -25.0 13
14 Six Targets in PondEx10 Field: SAS Images Images normalized to max pixel value in left image: 0 to 30 db range. 152 mm TP-T round (top) Small Al cylinder with notch Steel artillery shell Al replica of artillery shell 81 mm mortar (filled with cement) 155 mm howitzer shell (bottom) Cross Range (m) 5.5 3.5 1.5-0.5-2.5-4.5 0 20 40 60 80 9 10 11 Range (m) 14
15 Pulse Compressed Scattered Signals from 6 Targets Images normalized to max pixel value in left image: 0 to 30 db range. 0 20 40 60 80 152 mm TP-T round (top) Small Al cylinder with notch Steel artillery shell Al replica of artillery shell 81 mm mortar (cement filled) 155 mm howitzer shell (bottom) 18.25 m 3 ms 15
16 Acoustic Template after SAS Filtering 0 155 mm Howitzer Shell 152 mm TP-T Round 81 mm Mortar 0 90-6.3 Aspect Angle (deg.) 180-12.5 db (Relative) 270-18.3 360 0 8 16 24 32 f (khz) 0 8 16 24 32 f (khz) 0 8 16 24 32 f (khz) -25.0 16
17 Pulse Compressed Bistatic Signals from 6 Targets 0 20 40 Images normalized to max pixel value in each image: 0 to 30 db range. 152 mm TP-T round (top) Small Al cylinder with notch Steel artillery shell Al replica of artillery shell 81 mm mortar (cement filled) 155 mm howitzer shell (bottom) 60 80 targets 15 ms 18.25 m 17
18 30-50 khz Bistatic SAS Image 152 mm TP-T round (bottom) Small aluminum cylinder with notch Solid steel artillery shell Machined aluminum replica of artillery shell 81 mm mortar (filled with cement) 155 mm empty howitzer projectile (top) Targets rotated by 40 with respect to the STSM 2 rail system A Guide line at 11 m range 1 2 3 4 5 B NSWC PCD Sonar Tower 3 m Smoothed Areas 10 m Direction of Sonar Beam 21 m long STMS 2 Rail System 18
19 Data-Model Comparison for Solid Aluminum Cylinder Comsol FEM result via AxiScat Ver 1.1model from NURC Features identified using physical acoustics ray models Normalized Acoustic Templates Face crossing Rayleigh Wave PondEx09 Comsol FEM θ Meridional Rayleigh Wave γ a Broadside orientation at 0 End-on orientation at 90 L 19
20 Current State of Finite Element Models FE models results for free field and proud solid aluminum cylinder have been compared to data. Williams et al, J. Acoust. Soc. Am., 127, 3356-3371 (2010). Under an ONR post-doc, Dr. España has modified the model of Williams et al to simulate free field and proud scattering from an aluminum pipe. Computational meshes for the steel artillery shell and its aluminum and steel replicas have been developed with and without the surface feature (i.e., v - shaped grooves). Computational domain Uniform 1 cm mesh Refined around corners to 1 mm 20
21 Current State of Finite Element Models FE Model - data comparison for an aluminum pipe (2 ft long, 1 ft diameter, 3/8 inch wall thickness) suspended in the free field. Finite Element Model Data Measured at NSWC PCD 21
22 Current State of Finite Element Models FE Model - data comparison for an aluminum pipe (2 ft long, 1 ft diameter, 3/8 inch wall thickness) lying proud on a flat water-sand sediment interface. Finite Element Model PondEx10 Data 22
23 Class Separation of UXO Using Time-Frequency Features Target 1 Target 2 Target 3 Target 5 Slant Range 2 4 6 8 100 ROI of Target #1 Deep Blue Chosen Ping in Red frequency 1 0.8 0.6 0.4 Time Frequency Distribution of Segment of Ping within ROI 122 0.2 144 200 400 600 800 Aspect 0 0 0.5 1 1.5 2 time x 10-3 1 Ping Segment Highlights 1 Composite Highlights of 8 Pings 0.8 0.8 frequency 0.6 0.4 frequency 0.6 0.4 0.2 0.2 0 0 0.5 1 1.5 2 time x 10-3 0 0 0.5 1 1.5 2 time x 10-3 23
24 Class Separation of UXO Using Time-Frequency Features Percent of Total match 120 100 80 60 40 Classification of Targets Based on Random Sample of High Energy Returns template 1 template 2 template 3 template 4 template 5 20 0 1 2 3 4 5 Target # 24
25 Class Separation of Cylindrical Targets Using Features Processed buried target data Aspect = -27 o Physical interpretation of structure: helical wave 0.35 Unique physics-based features deduced from spectra and knowledge of target length Results fed into spectral algorithm that identifies peaks associated with the observed phenomenon 4 cylinders with 2:1 aspect ratio Log Amplitude 0.3 0.25 0.2 0.15 0.1 0.05 Blue: Data at -27 o aspect Red: Identified peaks by algorithm which looks for nearly harmonically related peaks Create feature vectors consisting of 5 parameters: number of quasi-harmonic peaks, start frequency, separation between peaks, a salient factor (value dependent on relative size of peaks relative to the background), aspect angle. 0 0 5 10 15 Dimensionless Frequency 25
26 Class Separation of Cylindrical Targets Using Features Process for using elastic information in sonar data to discriminate between 4 cylindrical targets with the same size and shape. Select pings based on minimum number of peaks in quasi-harmonic train and salience of train Create feature vectors from selected pings Cluster vectors using K means algorithm with N > T number of clusters Down select clusters based on influence of each target Obtain physics-based understanding of features in selected clusters Decrease N Silhouette plot of four clusters in 5 dimensional feature space demonstrate good class separation using chosen feature vector in a k-means algorithm. If the silhouette value is close to 1, it means that the sample is well-clustered and it was assigned to an appropriate cluster. 26
27 Summary Monostatic and bistatic acoustic scattering data sets over a wide frequency range have been processed with standard SAS imaging algorithms and acoustic templates have been generated. Targets included 6 UXO, 3 cylindrical targets, and 2 rocks. The SAS filtering algorithm has been found to be a robust means for isolating the scattered signal from an UXO in a cluttered environment. SAS images and acoustic templates suggest rich elastic responses are present on the UXO and these may be used in detection and classification algorithm. On-going finite element simulations are compared and validated against data. Bistatic SAS images have been created, but the data sets have yet to be exploited in a classification scenario such as acoustics templates. 27
28 Thank you for your attention. 28