PROBABILITY CHARACTERISTICS OF SENSOR OUTPUT DATA,

Abstract

The first results of a program to determine the statistical nature of surveillance sensor output data are presented. Most of the data was obtained from negative transparencies of vertically oriented aerial photographs. Smaller samples of radar and infrared films are reported for comparison. A line-scan process was used to generate an electrical analog of the light transmitted by the film. This signal was sampled, amplitude quantized, and recorded for later statistical analysis by a digital computer. Four statistical parameters were tabulated; specifically, the probability of occurrence of (a) quantized signal amplitudes, (b) quantized slopes between adjacent samples, (c) runs of constant amplitude, and (d) runs of constant slope. Application of the data is discussed in terms of coding and possible improvements in signal processing. Of the four characteristics, only slopes have probability distributions which permit significant reductions (a factor of 2 to 3) in channel capacity. Recommended extensions of the program are related to advanced sensors. (Author)

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1963
Accession Number
AD0418179

Entities

People

  • George W. Mcclure
  • John C. Dute

Organizations

  • University of Michigan

Tags

Communities of Interest

  • Sensors

DTIC Thesaurus Topics

  • Aerial Photographs
  • Amplitude
  • Channel Capacity
  • Computer Programming
  • Computers
  • Digital Computers
  • Infrared Film
  • Photographic Materials
  • Photographs
  • Photography
  • Probability
  • Probability Distributions
  • Signal Processing
  • Statistical Analysis

Readers

  • Approximation Theory.
  • Image Processing and Computer Vision.