The Scale-Space Formulation of Pyramid Data Structures.

Abstract

Pyramid data structures for image processing are usually defined using discrete grids and discrete levels. It has proven useful to formulate pyramids in terms of continuous variable. When the level of the pyramid is changed to be a continuous variables, we talk of the resulting domain as 'scale-space.' When both the image domain and level are treated as continuous, the resulting pyramid structures are most naturally viewed in terms of partial differential equations governing their formation. This viewpoint allows one to generalize to new kinds of pyramid data structures, analyze their information content, and develop rational methods for treating borders and other problems in the discrete construction of pyramids. Keywords: Gaussian and Laplacian pyramids; Zero crossings; Heat equation. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 1986
Accession Number
ADA178960

Entities

People

  • Robert A. Hummel

Organizations

  • New York University

Tags

Communities of Interest

  • Air Platforms
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algebraic Geometry
  • Analytic Functions
  • Boundaries
  • Boundary Value Problems
  • Computer Science
  • Construction
  • Convolution
  • Crossings
  • Differential Equations
  • Equations
  • Gaussian Distributions
  • Image Processing
  • Mathematical Analysis
  • Military Research
  • New York
  • Partial Differential Equations
  • Theorems

Readers

  • Calculus or Mathematical Analysis
  • Electromagnetic Wave Scattering and Antenna Radiation Engineering

Technology Areas

  • Space