Presentation is loading. Please wait.

Presentation is loading. Please wait.

CENTER FOR SUBSURFACE SENSING AND IMAGING SYSTEMS April 19, 2007 2007 CenSSIS Site Visit April 19, 2007 2007 CenSSIS Site Visit Miguel Vélez-Reyes R2C.

Similar presentations


Presentation on theme: "CENTER FOR SUBSURFACE SENSING AND IMAGING SYSTEMS April 19, 2007 2007 CenSSIS Site Visit April 19, 2007 2007 CenSSIS Site Visit Miguel Vélez-Reyes R2C."— Presentation transcript:

1 CENTER FOR SUBSURFACE SENSING AND IMAGING SYSTEMS April 19, 2007 2007 CenSSIS Site Visit April 19, 2007 2007 CenSSIS Site Visit Miguel Vélez-Reyes R2C Sub-thrust Leader Multi-Spectral Discrimination (MSD) Probe Multi-Band Detectors

2 Spectral Sensing and Imaging @ CenSSIS Detectors at different wavelengths, Y i Detectors at different wavelengths, Y i object Medium Clutter Broadband Probe, P Broadband Probe, P Remote Sensing Elastic-Scattering Spectroscopy Raman Imaging Spectroscopy

3 Spectral Sampling

4 Goals of Spectral Sensing & Imaging (R2C) Estimation, Detection, Classification, or Understanding o Crop health o Chemical composition, pH, CO 2 o Metabolic information o Ion concentration o Physiological changes (e.g., oxygenation) o Extrinsic markers (dyes, chemical tags) Examples of  Detect : presence of a target characterized by its spectral features  or  Classify: objects based on features exhibited in  or  Understand : object information, e.g., shape or other features based on  or . Integrating spatial and spectral domains. Or Estimate: probed spectral signature {  ( x,y, )} physical parameter to be estimated {  ( x,y, )}   M

5 MSSI Research Across Thrusts R2: Multispectral Physics-Based Signal Processing Fundamental Science Fundamental Science Validating TestBEDs Validating TestBEDs L1 L2 L3 S4 Bio-MedEnviro-Civil R3: Algorithm Implementation Benthic Habitat Mapping R1: Multispectral Imaging S1 Microscopy, Celular Imaging

6 R2C Research Work and Related R2 Image Representation –Scale space representation of hyperspectral imagery Image enhancement –Denoising of Raman Spectroscopy Signals Unmixing –Unsupervised unmixing using the constrained Positive Matrix Factorization –Subsurface unmixing Pattern Recognition –Subsurface detection and classification algorithms –Integration of spectral and spatial domains Change detection –Interest Operators –Multiband Adaptive Semiparametric Change Detection Image Understanding –Segmentation of Hyperspectral Imagery

7 Posters R2C –R2C p1: Tianchen Shi, Charles DiMarzio (NU), Multi-Spectral Reflectance Confocal Microscopy on Skin –R2C p6: Sol Cruz-Rivera, Vidya Manian (UPRM), Charles DiMarzio (NU), Component Extraction from CRM Skin Images –R2C p2: Melissa Romeo, Max Diem (NU), Vibrational Multispectral Imaging of Cells and Tissue: Monitoring Disease and Cellular Activity –R2C p3: Luis A. Quintero, Shawn Hunt (UPRM), Max Diem (NU), Denoising of Raman Spectroscopy Signals –R2C p4: Julio Martin Duarte-Carvajalino, Miguel Velez-Reyes (UPRM), Guillermo Sapiro (UM) Fast Multi-Scale Regularization and Segmentation of Hyperspectral Imagery via Anisotropic Diffusion and Algebraic Multigrid Solvers –R2C p5: Enid M. Alvira, Miguel Velez-Reyes, Samuel Rosario (UPRM) A Study on the Effect of Spectral Signature Enhancement in Hyperspectral Image Unmixing SeaBED –Sea p1: James Goodman, SeaBED: A Controlled Laboratory and Field Test Environment for the Validation of Coastal Hyperspectral Image Analysis Algorithms –Sea p2: Carmen Zayas, Spectral Libraries of Submerged Biotoped for Benthic Mapping in Southwestern Puerto Rico

8 Denoising of Raman Spectroscopy Signals: L. Quintero, S. Hunt, M. Diem Impulsive Noise Filter Savitzky-Golay Filter (Smoothing) Median Filter 7 point window Low pass Filter Cosmic Spike Classification |y[n]-u[n]|>thr Missing Point Filter + _ thr indx Cosmic Spikes Detection Wavelet Denoising + + + + Figure 1. Signal processing system: Impulsive noise filter and two alternatives to reduce the remaining noise (ν R [n]) Figure 2. Real spectra in blue and filtered signal in red using the impulsive noise filter Figure 3. Synthetic spectrum with Poisson noise. Estimations of s[n] using the Savitzky-Golay algorithm and Wavelets Shrinkage Estimators

9 Multi-Spectral Reflectance Confocal Microscopy on Skin: T. Shi, C. DiMarzio A new multi-spectral reflectance confocal microscopy to achieve sub- celluar functional imaging in skin by utilizing our unique Keck multi- modality microscope is presented. Ex-vivo and phantom experimental results are presented. Further development of this new modality may lead to future clinical applications.

10 Component Extraction from CRM Images S.M. Cruz-Rivera, V. Manian, C. DiMarzio Statistical techniques have been applied to extract components (endmembers) from CRM images of the skin. The results are compared with N-FINDR method of pure pixel extraction. Figure below shows the first 4 components from the ICA algorithm for wavelenght of 810nm. One image from the Original 4-D matrix ICA Results for CRM data for w = 810 nm Future work will include, spatial processing for extracting regional features and semi- supervised methods will be implemented to perform endmember extraction

11 Fast Multi-Scale Regularization and Segmentation of Hyperspectral Imagery via Anisotropic Diffusion and Algebraic Multigrid Solvers Grid 0 Grid S Grid s............ V-cycle Grid 0 Relax Grid S, Solve: Restriction Prolongation Grid s E : error, R: residual, V: approximated solution Julio M. Duarte (UPRM) Miguel Velez-Reyes (UPRM) Guillermo Sapiro (UMN)

12 A Study on the Effect of Spectral Signature Enhancement in Hyperspectral Image Unmixing Enid M. Alvira Miguel Vélez-Reyes Samuel Rosario Resolution Enhancement PCA Filter Unmixing

13 SeaBED: Sea p1 CONCEPT: Assemble a multi-level array of optical measurements, field observations and remote sensing imagery describing a natural reef system OBJECTIVE: Provide researchers with data from a fully-characterized test environment for the development and validation of subsurface aquatic remote sensing algorithms LEGACY: Utilize scientific publications and web-based distribution to establish Enrique Reef and its associated data as a lasting standard for algorithm assessment Benthic MeasurementsWater Column MeasurementsSurface Measurements Hyperspectral Image Data UPRM Researchers: J. Goodman, M. Vélez-Reyes, F. Gilbes, S. Hunt, R. Armstrong

14 SeaBED: Image Collection Campaign in Preparation, Sea p1

15 SeaBED: Spectral Library for Algorithm Validation Sea p2 New instrumentation and sampling techniques are being used for the development of spectral libraries required for hyperspectral subsurface unmixing algorithms.

16 Related Posters R1A –R1A p1: D. Goode, B. Saleh, A. Sergienko, M. Teich, Quantum Optical Coherence Tomography –R1A p2: A. Stern, O. Minaeva, N. Mohan, A. Sergienko, B. Saleh, M. Teich, Superconducting Single-Photon Dectectors (SSPDs) for OCT and QOCT –R1A p7: M. Dogan, J. Dupuis, A. Swan, Selim Unlu, B. Goldberg, Probing DNA on Surfaces Using Optical Interference Techniques R2B –R2B p3: Amit Mukherjee, Badri Roysam, Interest-points for Hyperspectral Images R2D –R2D p8: Karin Griffis, Maja Bystrom, Automatic Object-Level Change Interpretation for Multispectral Remote Sensing Imagery R3A –R3A p5: Carolina Gerardino, Wilson Rivera, James Goodman, Utilizing High- Performance Computing to Investigate Performance and Sensitivity of an Inversion Model for Hyperspectral Remote Sensing of Shallow Coral Ecosystems R3B –R3B p6: Samuel Rosario-Torres, Miguel Velez-Reyes, New Developments and Application of the MATLAB Hyperspectral Image Analysis Toolbox


Download ppt "CENTER FOR SUBSURFACE SENSING AND IMAGING SYSTEMS April 19, 2007 2007 CenSSIS Site Visit April 19, 2007 2007 CenSSIS Site Visit Miguel Vélez-Reyes R2C."

Similar presentations


Ads by Google