Feasibility of Optical Instruments Based on Multiaperture Optics.

Abstract

This paper deals with a recognition system specialized for low pixel numbers. It is assumed that the image consists of 15 x 15 pixels. Using such a low pixel number, the object is distorted beyond recognition if its straight edges are not lined up with the grid. This is especially important if the previously described method based on interpretation of each pixel row as a binary number is used. Therefore, it is shown that a given image of a randomly oriented object can be rotated by a subroutine requiring only little processing time. The resultant image deviates only to a minute degree from the image which would have been observed with its straight edges lined up with the pixel grid. Originator-supplied key words included: Low Pixel Numbers, Pattern Recognition, Multiaperture Optics, Recognition Coefficient.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 16, 1984
Accession Number
ADA150868

Entities

People

  • J. D. Cox
  • J. H. Jamed
  • R. T. Schneider

Organizations

  • University of Florida

Tags

DTIC Thesaurus Topics

  • Computer Graphics
  • Computers
  • Contracts
  • Detectors
  • Digital Image Processing
  • Digital Images
  • Distortion
  • Engineering
  • Graphics
  • Image Processing
  • Images
  • Language
  • Nuclear Engineering
  • Optical Instruments
  • Pattern Recognition
  • Recognition
  • Security

Fields of Study

  • Physics

Readers

  • Graph Algorithms and Convex Optimization.
  • Human-Computer Interaction (HCI).
  • Statistical inference.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference