Multidimensional Signal Coding in the Visual System

Abstract

This work shows that units in the visual system are tuned to principal components of real world color natural images and provides insight on the strategy the visual system may be using to confront natural imagery. It is also useful for identifying image components that are significant for the visual system and those which it either keeps or enhances for later stages of signal representation and analysis. The methods by which retinal center/surround R/G type cells can be used to realize this coding system are also investigated. In collaboration with NASA, the properties of a number of natural spatio-temporal sequences was measured. We found that the sequences showed represents .98 of the signal energy as separable. In addition, we developed an efficient coding model for natural time varying imagery in the early visual system. We show that M and P channels conjoin with tracking eye-movement to make visual system retinal architecture matched to basic components of time varying imagery.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 30, 1991
Accession Number
ADA244281

Entities

People

  • Gershon Buchsbaum

Organizations

  • University of Pennsylvania

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Air Force
  • Coding
  • Coefficients
  • Color Vision
  • Computer Programming
  • Conversion
  • Eigenvectors
  • Eye
  • Eye Movements
  • Filters
  • Frequency
  • Image Processing
  • Sequences
  • Signal Processing
  • Spectra
  • Three Dimensional
  • Universities

Fields of Study

  • Computer science

Readers

  • Computer Vision.
  • Systems Analysis and Design
  • Vision Science/Vision Psychology/Cognitive Neuroscience.