research thrust r3 image and data information management · leader, rpi john proakis co-leader, nu...
TRANSCRIPT
Research Thrust R3Image and Data Information Management
Research Thrust R3Image and Data Information Management
Diverse Problems – Similar SolutionsDiverse Problems – Similar Solutions
James ModestinoDavid Kaeli
James ModestinoDavid Kaeli
A. OverviewB. R3 Fundamental Research
R3a. Data Compression
R3b. Multimedia Networks
R3c. Scalable Computing
R3d. Image Databases and
Metadata
C. Impact and TimelineD. I-PLUS
A. OverviewB. R3 Fundamental Research
R3a. Data Compression
R3b. Multimedia Networks
R3c. Scalable Computing
R3d. Image Databases and
Metadata
C. Impact and TimelineD. I-PLUS
A. OverviewA. Overview
§ Acquisition, transmission, storage, processing, retrieval, dissemination, display and visualization of sensor data§ Development of a distributed collaborative research environment§ Provide remote Web access to I-PLUS resources
ScopeScope
SubsurfaceSensing and
Modeling
Validation Testbeds
Environmental-CivilApplications
Bio-MedicalApplications
Image and DataInformation
Management
Physics-BasedSignal Processing andImage Understanding
I–PLUS
R1R1R2R2
R3R3L1L1
L2L2
L3L3
ScopeScope
EngineeredSystemEngineeredSystem
EnablingTechnologiesEnablingTechnologies
FundamentalScienceFundamentalScience
§ Provide compression, processing, transmission and storage resources§ Provide communications/networking support infra-
structure and collaborative work environment
§ Provide compression, processing, transmission and storage resources§ Provide communications/networking support infra-
structure and collaborative work environment
§ Provide repository for tools and Solutionware§ Administer CenSSIS information management§ Provide Web access to CenSSIS resources
§ Provide repository for tools and Solutionware§ Administer CenSSIS information management§ Provide Web access to CenSSIS resources
L1L1
L2L2
L3L3
§ Develop efficient, flexible compression approaches§ Develop efficient/reliable transport schemes§ Develop new collaborative work environments§ Develop specific scalable computing approaches § Develop image database, archiving/search schemes
§ Develop efficient, flexible compression approaches§ Develop efficient/reliable transport schemes§ Develop new collaborative work environments§ Develop specific scalable computing approaches § Develop image database, archiving/search schemes
EE § Enable distributed education component§ Enhance engineering curriculum§ Enable distributed education component§ Enhance engineering curriculum
R3 GoalsR3 Goals
EngineeredSystemEngineeredSystem
EnablingTechnologiesEnablingTechnologies
FundamentalScienceFundamentalScience
EducationEducation
Focused orPulsed Probe
Focused orGated Detector
(LPM)
2D Photomosaic ofUnderwater Thermal Vent
WHOI Data Set
2D Photomosaic ofUnderwater Thermal Vent
WHOI Data Set
§ Processing/Storage/Transmission Problem§ Transmission Time 40 min @ 100 Mb/s§ Compression Reduces to < 0.5 min§ Layered Progressive Scheme Useful
§ Processing/Storage/Transmission Problem§ Transmission Time 40 min @ 100 Mb/s§ Compression Reduces to < 0.5 min§ Layered Progressive Scheme Useful
MultipleOverlapping
MultipleOverlapping
OpticalImagesOpticalImages
SensorData
SensorData
Typical Localized Probing & Mosaicing(LPM) ApplicationTypical Localized Probing & Mosaicing(LPM) Application
104 1K x 1K 24 30 GBytes
Number ofImages
Number ofImages
Image SizeIn Pixels
Image SizeIn Pixels
Number ofBits/PixelsNumber ofBits/Pixels
Total SizeOf Data SetTotal Size
Of Data Set
Single Quadrature Microscope
View of Mouse EggMultiple Views perSlice and Multiple
Slices per ReconstructionMultiple Reconstructions
for Moving ObjectsNU/MGH Data Set
Single Quadrature Microscope
View of Mouse EggMultiple Views perSlice and Multiple
Slices per ReconstructionMultiple Reconstructions
for Moving ObjectsNU/MGH Data Set
MultipleViews
MultipleViews
PossibleTemporal
Sequences
PossibleTemporal
Sequences SensorData
SensorData
Object
Detectors
Sources
(MVT)
§ Transmission Time 4 min @ 100 Mb/s§ For 100 Reconstructions is 400 min§ Compression Reduces to < 1 min§ Utilize 3-D Volumetric Compression§ Multiresolution Layered Scheme Useful
§ Transmission Time 4 min @ 100 Mb/s§ For 100 Reconstructions is 400 min§ Compression Reduces to < 1 min§ Utilize 3-D Volumetric Compression§ Multiresolution Layered Scheme Useful
Number ofViews/Slices
10
Number ofViews/Slices
10
Image SizeIn Pixels/View
1K x 1K
Image SizeIn Pixels/View
1K x 1K
Number ofBits/Pixels
24
Number ofBits/Pixels
24
Total SizeOf Data Set3 GBytes
Total SizeOf Data Set3 GBytes
Number ofSlices
100
Number ofSlices
100
Typical Multi-View Tomography (MVT) ApplicationTypical Multi-View Tomography (MVT) Application
Remote HyperspectralSensing of Coastal Regions
UPRM Data Set
Remote HyperspectralSensing of Coastal Regions
UPRM Data Set
MultipleWavelengths
MultipleWavelengths
MultipleImagesMultipleImages
SensorData
SensorData
(MSD)
Wide BandProbe
Narrow BandDetectors
backgroundagua1agua-2agua4basuracultivo1cultivo2cultivo-recortadomontanascultivo4cutlivo-costarenazona-urbanaagua-5cultivo6agua6techo-almacencultivo-corto-Nagua-costa-Nvegetacionrio
Number ofImages/
Wavelength103
Number ofImages/
Wavelength103
Image SizeIn Pixels/View
1K x 1K
Image SizeIn Pixels/View
1K x 1K
Number ofBits/Pixels
16
Number ofBits/Pixels
16
Total SizeOf Data Set20 TBytes
Total SizeOf Data Set20 TBytes
Number ofWavelengths
103
Number ofWavelengths
103
§ Transmission Time 56 hr @ 100 Mb/s§ Compression Reduces to < 30 min§ Exploit Correlation Across Wavelengths§ Layered Progressive Scheme Useful
§ Transmission Time 56 hr @ 100 Mb/s§ Compression Reduces to < 30 min§ Exploit Correlation Across Wavelengths§ Layered Progressive Scheme Useful
Typical Multi-Spectral Discrimination (MSD) ApplicationTypical Multi-Spectral Discrimination (MSD) Application
Focused orPulsed Probe
Focused orGated Detector
(LPM) (MSD)
Wide BandProbe
Narrow BandDetectors
Photomosaic of the Curved Human Retina
RPI Data Set
Photomosaic of the Curved Human Retina
RPI Data Set
MultipleOverlapping
MultipleOverlapping
Temporal Sequence of
Multi-SpectralImages
Temporal Sequence of
Multi-SpectralImages Sensor
DataSensor
Data
A: Retinal Surface (visible channel)B: Subsurface (near infrared)
A: Retinal Surface (visible channel)B: Subsurface (near infrared)
§ Real-Time Transmission Requires 720Mb/s § Compression Reduces to < 6 Mb/s§ Transmission Time 6.6 hr @ 100 Mb/s§ Layered Transport Scheme Useful
§ Real-Time Transmission Requires 720Mb/s § Compression Reduces to < 6 Mb/s§ Transmission Time 6.6 hr @ 100 Mb/s§ Layered Transport Scheme Useful
Image SizeIn Pixels1K x 1K
Image SizeIn Pixels1K x 1K
Number ofBits/Pixels
12
Number ofBits/Pixels
12
Total SizeOf Data Set
300-600 GBytes
Total SizeOf Data Set
300-600 GBytes
Length of Procedure
1-2 hrs
Length of Procedure
1-2 hrs
Number ofChannels
2
Number ofChannels
2Frame Rate
30/secFrame Rate
30/sec
Typical LPM/MSD ApplicationTypical LPM/MSD Application
How Can We Provide Adequate Communications/Networking Infrastructure?How Can We Provide Adequate Communications/Networking Infrastructure?
CenSSISCenSSIS
backgroundagua1agua-2agua4basuracultivo1cultivo2cultivo-recortadomontanascultivo4cutlivo-costarenazona-urbanaagua-5cultivo6agua6techo-almacencultivo-corto-Nagua-costa-Nvegetacionrio
ClustersClusters CollaborationCollaboration
ConferencingConferencing
Web AccessWeb Access
DatabasesDatabases
Multimedia NetworkInternet 2
Multimedia NetworkInternet 2
TestbedsTestbeds
B. Fundamental ResearchB. Fundamental Research
Goal: Develop information management concepts, tools and infrastructure to support basic CenSSIS scientific mission
Goal: Develop information management concepts, tools and infrastructure to support basic CenSSIS scientific mission
§ Distribution
§ Transport
§ Collaboration
§ Web Access
§ Distribution
§ Transport
§ Collaboration
§ Web Access
§ Representation
§ Compression
§ Transmission
§ Storage
§ Representation
§ Compression
§ Transmission
§ Storage
§ Computing
§ Storage
§ Scalability
§ Solutionware
§ Computing
§ Storage
§ Scalability
§ Solutionware
§ Archiving
§ Image Formatting
§ Image Browsing
§ Data Migration
§ Archiving
§ Image Formatting
§ Image Browsing
§ Data Migration
Components of R3 Research ThrustComponents of R3 Research Thrust
R3Image and Data
InformationManagement
R3Image and Data
InformationManagement
R3dImage Databases
and Metadata
R3dImage Databases
and Metadata
R3cScalable
Computation
R3cScalable
Computation
R3bMultimedia Networks
R3bMultimedia Networks
R3aData
Compression
R3aData
Compression
R3a/b Data Compression/
Multimedia Networks
R3a/b Data Compression/
Multimedia Networks
R3c/d Scalable Computation/Database and Metadata
R3c/d Scalable Computation/Database and Metadata
James ModestinoLeader, RPI
John ProakisCo-Leader, NU
Shiv Kalyanaraman, RPIMasoud Salehi, NUJohn Woods, RPIGary Saulnier, RPIGreg Suski, LLNL
David KaeliLeader, NU
Robert FutrelleCo-Leader, NU
Jose Cruz-Rivera, UPRMRoscoe Giles, BUClaudio Rebbi, BUElias Manolakos, NUWaleed Meleis, NU
The R3 Research TeamThe R3 Research Team
Image/VideoSensor DataAcquisition
Image/VideoSensor DataAcquisition
PreprocessingRepresentationCompression
EC Coding
PreprocessingRepresentationCompression
EC Coding
User Applications
Render/EnhanceDisplay/Playback
User Applications
Render/EnhanceDisplay/Playback
EC DecodingDecompressionReconstructionPostprocessing
EC DecodingDecompressionReconstructionPostprocessing
Storage/ArchivingIndexingRetrieval
Storage/ArchivingIndexingRetrieval
ModulationTransmissionNetworkingDistribution
Demodulation
ModulationTransmissionNetworkingDistribution
Demodulation
Issues and Requirements Depend on Sensor Modality Type and Communications/Storage Infrastructure
Issues and Requirements Depend on Sensor Modality Type and Communications/Storage Infrastructure
TransmissionChannel
TransmissionChannel
StorageChannelStorageChannel
Object
MSDMSD
LPMLPM MVTMVT
R3a/b Image and Data Management IssuesR3a/b Image and Data Management Issues
§ Multi-resolution, or layered, coding approaches§ Model-based compression schemes§ Feature-based compression schemes§ Multi-objective compression schemes§ Data fusion in compressed domain§ Joint Source-Channel Coding (JSCC) schemes§ Applications built on enhanced TCP/IP networking
protocols§ Utilize existing lower-layer networking infrastructure
§ Multi-resolution, or layered, coding approaches§ Model-based compression schemes§ Feature-based compression schemes§ Multi-objective compression schemes§ Data fusion in compressed domain§ Joint Source-Channel Coding (JSCC) schemes§ Applications built on enhanced TCP/IP networking
protocols§ Utilize existing lower-layer networking infrastructure
R3a/b Compression, Transmission and StorageR3a/b Compression, Transmission and Storage
Barrier 3Barrier 3 Lack of Robust, Physics-Based Recognition and Sensor Fusion TechniquesLack of Robust, Physics-Based Recognition and Sensor Fusion Techniques
Barrier 6Barrier 6 Lack of Rapid Processing and Management of Large Image DatabasesLack of Rapid Processing and Management of Large Image Databases
R3a Compression Research IssuesR3a Compression Research Issues
Existing Approaches§ General Purpose§ Intended for Human
Evaluation§ Based on HVS Distortion
Criterion§ Generally Single Resolution
Lossy Compression§ Compression Efficiency
Saturated at < 100:1
Existing Approaches§ General Purpose§ Intended for Human
Evaluation§ Based on HVS Distortion
Criterion§ Generally Single Resolution
Lossy Compression§ Compression Efficiency
Saturated at < 100:1
CenSSIS Approach§ Physics-Based§ Intended for Specific
Imaging Functions§ Based on Function-Specific
Distortion Criterion§ Provides Progressive
Lossy-LosslessCompression§ Compression Efficiency
Improvements by 10-100
CenSSIS Approach§ Physics-Based§ Intended for Specific
Imaging Functions§ Based on Function-Specific
Distortion Criterion§ Provides Progressive
Lossy-LosslessCompression§ Compression Efficiency
Improvements by 10-100
§ Will Leverage Existing Approaches Where and When Possible§ Significant Improvements in Compression Efficiency
Critical to Success of CenSSIS Scientific Mission
Physics-Motivated Region-Based CodingPhysics-Motivated Region-Based Coding
Original-HealthyOriginal-Healthy
RegionsRegions
§ Macula/Fovea Region-Center§ Optical Disc-Left§ Vascular Structure § Background
§ Macula/Fovea Region-Center§ Optical Disc-Left§ Vascular Structure § Background
SegmentationSegmentation
ROI CodingROI Coding
§ Macula Region-Near Lossless§ Optical Disc-Fine§ Vascular Structure-Medium § Background-Coarse
§ Macula Region-Near Lossless§ Optical Disc-Fine§ Vascular Structure-Medium § Background-Coarse
Multiresolution ApproachMultiresolution Approach
Dyadic 3-LevelWavelet SubbandDecomposition
Dyadic 3-LevelWavelet SubbandDecomposition
FullbandFullband
HalfBandHalfBand
LowbandLowband
H-HH-H
H-LH-L
L-HL-H
LL-LLLL-LL
LH-LHLH-LHLL-LHLL-LH
LH-LLLH-LL
L - Low Frequency ComponentH - High Frequency ComponentL - Low Frequency ComponentH - High Frequency Component
SchematicRepresentationSchematicRepresentation
§ Coarse segmentation at lowest band§ Use physics-motivated segmentation§ Successive refinement with higher bands§ Differential ROI coding of regions§ Coding approach tailored to region type§ Exploit correlation across scales§ Progressive Lossy-Lossles encoding
§ Coarse segmentation at lowest band§ Use physics-motivated segmentation§ Successive refinement with higher bands§ Differential ROI coding of regions§ Coding approach tailored to region type§ Exploit correlation across scales§ Progressive Lossy-Lossles encoding
Operational UseOperational Use
Original-HealthyOriginal-Healthy Original-DiseasedOriginal-Diseased
§ Presence of Pathology Obscures Underlying Region Types § How to Reliably Extract Important Regions with Pathology§ How to Exploit the Physical Manifestations of Disease
Pathologies
§ Investigate and assess compression, transmission and storage requirements associated with initial level L2 testbeds
§ Develop preliminary modality-specific compression Approaches:§ LPM§ MVT§ MSD
§ Provide preliminary extensible collaborative work environment based on IP telephony model
§ Develop transition plan to Internet-2 and identify and formulate middleware and QoS research issues
§ Identify and demonstrate promising shared applications such as Virtual Lab
§ Investigate and assess compression, transmission and storage requirements associated with initial level L2 testbeds
§ Develop preliminary modality-specific compression Approaches:§ LPM§ MVT§ MSD
§ Provide preliminary extensible collaborative work environment based on IP telephony model
§ Develop transition plan to Internet-2 and identify and formulate middleware and QoS research issues
§ Identify and demonstrate promising shared applications such as Virtual Lab
Object
MSDMSD
LPMLPM MVTMVT
R3a/b First Year Anticipated ResultsR3a/b First Year Anticipated Results
§ Distribution
§ Transport
§ Collaboration
§ Web Access
§ Distribution
§ Transport
§ Collaboration
§ Web Access
§ Representation
§ Compression
§ Transmission
§ Storage
§ Representation
§ Compression
§ Transmission
§ Storage
§ Computing
§ Storage
§ Scalability
§ Solutionware
§ Computing
§ Storage
§ Scalability
§ Solutionware
§ Archiving
§ Image Formatting
§ Image Browsing
§ Data Migration
§ Archiving
§ Image Formatting
§ Image Browsing
§ Data Migration
Components of R3 Research ThrustComponents of R3 Research Thrust
R3Image and Data
InformationManagement
R3Image and Data
InformationManagement
R3dImage Databases
and Metadata
R3dImage Databases
and Metadata
R3cScalable
Computation
R3cScalable
Computation
R3bMultimedia Networks
R3bMultimedia Networks
R3aData
Compression
R3aData
Compression
§ 4096-node system§ Enables spatial and
spectral slicing§ UPRM will develop parallelized
versions of CenSSIS algorithms for focal plane processing§ End-to-end hardware/software
sensor design
§ 4096-node system§ Enables spatial and
spectral slicing§ UPRM will develop parallelized
versions of CenSSIS algorithms for focal plane processing§ End-to-end hardware/software
sensor design
TodayToday FutureFutureImage ResolutionImage Resolution 1,000 x 1,0001,000 x 1,000 10,000 x 10,00010,000 x 10,000Dynamic RangeDynamic Range 10 bits / pixel10 bits / pixel 12 bits / pixel12 bits / pixel
Scanning WavelengthsScanning Wavelengths 200 samples200 samples 1000 samples1000 samplesFrame rateFrame rate 30 fps30 fps 30 fps30 fps
Required BandwidthRequired Bandwidth 60 Gbits/s60 Gbits/s 36 Tbits/s36 Tbits/s
SIMD Focal Plane ArchitectureSIMD Focal Plane Architecture
Georgia Tech/UPRM SIMPil Architecture
Georgia Tech/UPRM SIMPil Architecture
ACUACU
R3c. How Do We Meet the Real-Time Processing Requirements of Hyperspectral Imaging? R3c. How Do We Meet the Real-Time Processing Requirements of Hyperspectral Imaging?
R3c. How can CenSSIS Applications Scale Up to Use Real Data Sets?R3c. How can CenSSIS Applications Scale Up to Use Real Data Sets?
§ Exploit NSF Partnerships for Advanced Computational Infrastructure (PACI) resources§ Construct Beowulf-class clusters to provide low-
cost, scalable processing (BioMed 8-node cluster and NSF MRI 32-node cluster)§ Utilize local NSF PACI/GRID computational
resources at Boston University (Mariner Center)§ Develop parallel programming toolsets around
MPI and Open/MP standards
§ Exploit NSF Partnerships for Advanced Computational Infrastructure (PACI) resources§ Construct Beowulf-class clusters to provide low-
cost, scalable processing (BioMed 8-node cluster and NSF MRI 32-node cluster)§ Utilize local NSF PACI/GRID computational
resources at Boston University (Mariner Center)§ Develop parallel programming toolsets around
MPI and Open/MP standards
CCS Center for Computational Science
SCV Scientific Computing and Visualization Group
InstrumentsInstruments
AlgorithmsAlgorithms
ImagesImages
Experi-mental
and ModelingResults
Experi-mental
and ModelingResults
How Do We Store/Organize CenSSIS Information to Enable Rapid Assessment and Dissemination?How Do We Store/Organize CenSSIS Information to Enable Rapid Assessment and Dissemination?
y = C(x, z) + ny = C(x, z) + n
Database (Oracle8i)Database (Oracle8i)
Metadata - eXtensible Markup Language (XML)Metadata - eXtensible Markup Language (XML)
*TBs of RAID storage
*Replication and distribution
*TBs of RAID storage
*Replication and distribution
*Image format standards
*Image format standards
§ The data about the data§ Contextual browsing§ The data about the data§ Contextual browsing
R3d. CenSSIS DatabasesR3d. CenSSIS Databases
Image IDLPM099MVT002MSD321MVT061
Image typeHyperspectral
ECGDiffusive wave
CAT scan
Image format????
SensorCompression
algorithm
ManufacturerSoftwareversion
Calibration dateRelated
Publications
Serial #HLLs
supported
§ Need for Software Engineering standards§ Standard programming interfaces (e.g., APIs)§ Version and release control§ Testing standards
§ Enable rapid prototyping of new algorithms§ Develop parallel implementations for compute-
bound algorithms and models§ Provide libraries of Solutionware toolboxes,
validated in our L2 testbeds§ Leverage PACI and industrial-partner software
development tools
y = C(x, z) + nfor (i =0; i< n, i++)
SolutionwareSolutionwareAlgorithms and ModelsAlgorithms and Models
R3d. How do we develop the subsurface imaging toolbox of the future?R3d. How do we develop the subsurface imaging toolbox of the future?
R3c/d First Year Anticipated ResultsR3c/d First Year Anticipated Results
§ Complete benchmarking of SIMPiL focal plane architecture and applications§ Set up CenSSIS web, image storage and
database servers§ Design and implement CenSSIS image database
system§ Develop CenSSIS image format standards§ Provide preliminary specification for CenSSIS
metadata databases§ Develop CenSSIS software library control system§ Complete prototype of JavaPorts programming
toolset
§ Complete benchmarking of SIMPiL focal plane architecture and applications§ Set up CenSSIS web, image storage and
database servers§ Design and implement CenSSIS image database
system§ Develop CenSSIS image format standards§ Provide preliminary specification for CenSSIS
metadata databases§ Develop CenSSIS software library control system§ Complete prototype of JavaPorts programming
toolset
C. R3 Impact and TimelineC. R3 Impact and Timeline
§ Deliver experiments and testbeds to the classroom§ Immediate practice of classroom theory
§ Remote virtual laboratory*§ New Course modules§ Digital Image and Video Processing (ECSE-6630)§ Digital Picture Processing (ECSE-6640)§ Digital Computer Architecture (ECE-3391)
§ Collaborative work environment§ Distance education and joint seminars
§ Allow students to explore using CenSSIS Solutionwareand image databases§ Develop computer-based analysis skills applied to subsurface
sensing and imaging**
§ Deliver experiments and testbeds to the classroom§ Immediate practice of classroom theory
§ Remote virtual laboratory*§ New Course modules§ Digital Image and Video Processing (ECSE-6630)§ Digital Picture Processing (ECSE-6640)§ Digital Computer Architecture (ECE-3391)
§ Collaborative work environment§ Distance education and joint seminars
§ Allow students to explore using CenSSIS Solutionwareand image databases§ Develop computer-based analysis skills applied to subsurface
sensing and imaging**
R3 Impact on EducationR3 Impact on Education
*Already being piloted in RPI Nuclear Navy program
**Already being piloted in NU GE1103 course
*Already being piloted in RPI Nuclear Navy program
**Already being piloted in NU GE1103 course
ACUACU
R3 Technology: Real-time HyperspectralImagingR3 Technology: Real-time HyperspectralImaging
Barrier 5Barrier 5 Lack of Optimal End to End Sensor Design MethodsLack of Optimal End to End Sensor Design Methods
Lack of Validated, Integrated Processing and Computation ToolsLack of Validated, Integrated Processing and Computation ToolsBarrier 7Barrier 7
SolutionwareSolutionware
Inverse ECG speedup
02468
0 5 10Nodes
Sp
eed
up
R3 Technology: Scalable Inverse SolutionsR3 Technology: Scalable Inverse Solutions
y = C(x, z) + n for (i =0; i< n, i++)
Lack of Validated, Integrated Processing and Computation ToolsLack of Validated, Integrated Processing and Computation ToolsBarrier 7Barrier 7
Linear speedupLinear speedup
CenSSIS CenSSIS
R3 Technology: Commercial Databases, Compression and Multimedia NetworksR3 Technology: Commercial Databases, Compression and Multimedia Networks
Barrier 6Barrier 6 Lack of Rapid Processing and Management of Large Image DatabasesLack of Rapid Processing and Management of Large Image Databases
MultimediaNetwork
backgroundagua1agua-2agua4basuracultivo1cultivo2cultivo-recortadomontanascultivo4cutlivo-costarenazona-urbanaagua-5cultivo6agua6techo-almacencultivo-corto-Nagua-costa-Nvegetacionrio
Deliver image DB
Develop web-based query capabilities
Maintenance and upgrades to image database
Metadatadesign spec
Development/implementation of XML-based metadata database
Contextual browsing
Simulation of SIMPiL
Development of parallel hyperspectral applications
Evaluation of performance on SIMPiL hardware
Construct library system
Develop web-based rapid prototyping library system
Continued support and test of evolving Solutionware libraries
Assessment of L2 needs
New modality-specific compression approaches
Continued developmentfor new L2/L3 approaches
Assessment of L2 needs
Development of L2 support infrastructure based on
Internet-2Continued Investigation ofTransport and QoS Issues
Virtual lab applications
Extend shared applications to Internet-2 model
Continued Development Basedon Internet Evolution
Yr 5Yr 4Yr 3Yr 2Yr 1Education
Focal-plane architectures and algorithms
Image databases and metadata
Compression,Transmission and Storage
Multimedia Networking and Collaborative Work Environments
Solutionware
R3 TimelineR3 Timeline
SeaBEDSeaBED BioBEDBioBED
R1 ModelingR2 Info Extraction
R3 TechnologiesR3 Technologies
R3 TechnologiesR3 Technologies
CenSSISProductsCenSSISProducts
IndustryIndustry Academic ResearchAcademic Research
Less
ons l
earn
ed
R3 Integration into I-PLUSR3 Integration into I-PLUS
ImagesImages SW Libraries
SW Libraries SensorsSensors ToolsTools
MVT2MVT2LPM2LPM2MVT1MVT1 LPM1LPM1
D. I-PLUSD. I-PLUS
I-PLUSI-PLUS
§ I-PLUS§ Develops a unified problem-solving
platform§ Includes techniques, tools and infrastructure§ Combines the capabilities of the L2 testbeds
and Systems Integration § Allows CenSSIS to address new sets of
real-world problems§ Leveraging the combined resources, techniques
and lessons learned from multiple sensing and imaging modalities§ Provides a natural path for technology transfer
to outreach collaborators and industry partners
§ I-PLUS§ Develops a unified problem-solving
platform§ Includes techniques, tools and infrastructure§ Combines the capabilities of the L2 testbeds
and Systems Integration § Allows CenSSIS to address new sets of
real-world problems§ Leveraging the combined resources, techniques
and lessons learned from multiple sensing and imaging modalities§ Provides a natural path for technology transfer
to outreach collaborators and industry partners
CenSSIS Software Resource Center
§CenSSIS Software Engineer
§ Software libraries
§ Programming Tools and Parallel Systems Management
CenSSIS Software Resource Center
§CenSSIS Software Engineer
§ Software libraries
§ Programming Tools and Parallel Systems Management
CenSSIS Image and Data Information Systems
§CenSSIS Database Administrator
§Access to parallel processing resources
§Website Administration
CenSSIS Image and Data Information Systems
§CenSSIS Database Administrator
§Access to parallel processing resources
§Website Administration
I-PLUS - Integrated Probes to Look Under SurfacesI-PLUS - Integrated Probes to Look Under Surfaces
Systems Integration
D. Kaeli, NU
Leader
Systems Integration
D. Kaeli, NU
Leader