Multi-Signal, Multi-Modal Data Acquisition and Processing Based on Compressive Sensing

Abstract

In this final report, we report on our progress on ARO grant W911NF-07-1-0502. The goal of the project was to develop compressive sensing and dimensionality reduction for manifold data. We investigated and developed efficient sampling and measurement schemes for manifold-modeled data to enable efficient new A/AiTR and fusion algorithms. Key elements included random projections, dimensionality reduction techniques, and the new theory of compressive sensing. We developed new methods for signal and image registration, reconstruction, fusion, classification, and detection from incomplete information based on new compressive matched filters, MACH filters, and pattern recognition techniques. The research highlights detailed below are (i) a new smashed filter for dimensionally reduced classification and A/AiTR; (ii) new algorithms for machine and manifold learning based on random projections; (iii) joint manifold models and processing algorithms for multi-sensor and multi-modal data fusion.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 16, 2008
Accession Number
ADA501161

Entities

People

  • Richard G. Baraniuk

Organizations

  • Rice University

Tags

Communities of Interest

  • Autonomy
  • Human Systems
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Acquisition
  • Algorithms
  • Compressed Sensing
  • Data Acquisition
  • Data Processing
  • Dimensionality Reduction
  • Image Processing
  • Information Processing
  • Information Science
  • Information Systems
  • Information Theory
  • Machine Learning
  • Measurement
  • Signal Processing
  • Supervised Machine Learning
  • Target Recognition
  • Three Dimensional

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Image Processing and Computer Vision.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms