THE PRENORMALIZATION OF RECONNAISSANCE DATA.

Abstract

Several techniques are described and evaluated for the prenormalization of gray-scale aerial photographs prior to their use as input to an adaptive memory recognition system. It summarizes the evidence for analagous preprocessing operations in the animal visual system. Discussed are the mplementation and testing of several pre-processing techniques which reduce the gray-scale input data to binary form. The utility of candidate techniques in connection with automatic detection of targets on aerial photographs is estimated by comparing recognition error rates on a computersimulated adaptive memory recognition system. Problems associated with high-speed prenormalizing techniques were investigated through the construction of a real time gradient mapping system which also permits evaluation of certain parameter variations on various types of target images. A preferred implementation is described for a versatile prenormalization device, suitable for use in various recognition tasks. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jul 25, 1963
Accession Number
AD0422946

Entities

People

  • H. G. Kellett
  • J. S. Bryan
  • John F. Bogusz

Tags

DTIC Thesaurus Topics

  • Aerial Photographs
  • Automatic
  • Construction
  • Detection
  • Gray Scale
  • Images
  • Photographic Materials
  • Photographs
  • Photography
  • Preprocessing
  • Recognition
  • Reconnaissance
  • Test And Evaluation

Readers

  • Computer Vision.
  • Neural Network Machine Learning.