Multiscale Statistical Models for Signal and Image Processing

Abstract

We are developing a general theory for multi scale signal and image modeling, processing, and analysis that matched to singularity-rich data, such as transients and images with edges. Using a linguistic analogy, our model can be interpreted as grammars that constrain the wavelet vocabulary. Our investigation focuses on probabilistic graph models (tree-based hidden Markov models) that can accurately, realistically, and efficiently represent singularity structure in the wavelet domain. Grammar design is being guided by a detailed study of the final structure of singularities using Besov spaces and multifractal analysis.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 2004
Accession Number
ADA425177

Entities

People

  • Richard G. Baraniuk

Organizations

  • Rice University

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Abstracts
  • Algorithms
  • Classification
  • Coefficients
  • Geometry
  • Grammars
  • Hidden Markov Models
  • Image Compression
  • Image Processing
  • Markov Models
  • Models
  • Probability
  • Signal Processing
  • Students
  • Two Dimensional
  • Vocabulary
  • Wavelet Transforms

Readers

  • Computational Fluid Dynamics (CFD)
  • Graph Algorithms and Convex Optimization.
  • Image Processing and Computer Vision.

Technology Areas

  • Space