210 likes | 377 Vues
HSI is a Key Technology. Environmental monitoringNASA FloraCHRIS (Compact High Resolution Imaging Spectrometer)Proba (ESA), HERO (Canadian), SPECTRA (ESA), and EnMAP (German) missions.DoD Situational AwarenessAFRL/Raytheon TacSat 3 ARTEMISSpace ExplorationNASA MRO Compact Reconnaissance Im
E N D
1. Prof. Miguel Vélez-Reyes
Lab. for Appl. Remote Sensing and Image Proc.
Univ. of Puerto Rico at Mayaguez
S. Rosario-Torres,
J. Goodman, V. Manian Hyperspectral Image Exploitation
for Ship Detection
2. HSI is a Key Technology Environmental monitoring
NASA Flora
CHRIS (Compact High Resolution Imaging Spectrometer)
Proba (ESA),
HERO (Canadian),
SPECTRA (ESA), and
EnMAP (German) missions.
DoD Situational Awareness
AFRL/Raytheon TacSat 3 ARTEMIS
Space Exploration
NASA MRO Compact Reconnaissance Imaging Spectrometer for Mars (CRISM)
NASA Moon Mineral Mapper (M3) mission
3. The problem of interest
4. Challenges How to combine different modalities to optimize information extraction
Dynamic
High dimensionality
Variability introduced by
Changes in atmospheric conditions
Differences in illumination, orientation, etc
Variable unstructured clutter in standoff applications
Mixed signatures (clutter and threat)
5. Proposed Approach Powerful Methods for Constructing Detectors and Classifiers
Kernel-based methods
Support Vector Machines (SVM)
Adaptive Boosting Techniques
AdaBoost
Dimensionality Reduction andFeature Extraction
Invariant features
Optimize sensor combinations
Adaptation and Nonlinear Learning
Changing environment
Robust detection of new classes of targets
Optimize sensor combinations Explotion of new and powerful methods. Learn from the data differences between threat and non threat situations.
Novel methods for dimensionality reduction. Techniques that reduce the amount of data to transmit over sensor networks. Compresed sensing has shown that the information required to separate object classes with significant variability induced by chages in pose, illumination, etc.. Is summarized in few dimensions with techniques that are object indepentent.
Another challeging direction is the detection of new classes of explosives that appear as annomalous signature and the automated re-training of the classifier.Explotion of new and powerful methods. Learn from the data differences between threat and non threat situations.
Novel methods for dimensionality reduction. Techniques that reduce the amount of data to transmit over sensor networks. Compresed sensing has shown that the information required to separate object classes with significant variability induced by chages in pose, illumination, etc.. Is summarized in few dimensions with techniques that are object indepentent.
Another challeging direction is the detection of new classes of explosives that appear as annomalous signature and the automated re-training of the classifier.
6. Our Expertise Hyperspectral image processing
Vector/Multichannel image processing
Classification and detection in high dimensional feature spaces
Nonlinear Signal Processing
Machine learning
Automatic target recognition
7. Geometric PDE Processing of HSI: Object Oriente Approaches Improve Target Background Contrast
Improve Detection and Classification
8. Unsupervised Unmixing: Target Clutter Separation
9. Algorithm Implementation: Solutionware
12. Alternative Computational Platforms for Hyperspectral Image Processing Problem of Interest: Study alternative platforms where hyperspectral algorithms may be mapped efficiently,
Algorithm
Unsupervised unmixing
Platforms
Massively parallel processors – CUDA GPGPUs
Field programmable gate arrays - FPGAs
Features:
Embarrasingly parallel structure
Tune application to platforms.
18. 2007 Puerto Rico Hyperspectral Mission*
19. Space Information Laboratory Provides Satellite Reception Capabilities: Investment of approximately $1.3M in Infrastructure.
Only university under the U.S. flag licensed to receive LANDSAT 7.
Trains students in station operations, programming, image processing, satellite tracking, state of the art high data rate communications, and more.
S band station receives NOAA telemetry (12, 14, 15, 16, 17) and SeaWiFS data.
X-band station receives LANDSAT 7, RADARSAT 1 and MODIS (Terra and Aqua Satellite).
NASA FUSE Ground Control Station
SIL provides imagery for the other TCESS components.
20. Initial Focus Geometric PDEs for spectral/spatial integration for image segmentation
Hyperspectral Target Detection
Spectral Libraries (Collaboration with UH)
High spatial resolution HSI
Sub-pixel Target Detection
Analysis of MODIS/AVHRR Imagery over the Caribbean
21. Synergy with CIMES Collaborators UH:
Design and construction on hyperspectral airborne imagers
Hawaii Space Flight Lab access to space
UAF
Real time processing
Builds on UPRM’s expertise:
Over 10 years of work in the area
Over 100 peer reviewed publications
2 Book chapters
Research sponsored by: NSF, DoD, DHSAFRL, NOAA, NASA, NGA
22. Synergy with Existing Centers: DHS, NSF