Computer Estimated Probability of Detection: Can You Hide from a Computer?

Abstract

Several clutter metrics were evaluated and compared against the probability of detection of ground combat vehicle targets in test scenes created in a natural field environment. This paper presents methods of testing subjects and methods of computing the metrics. Finally, limited results of the initial testing and the comparison of this against the metrics are given, thereby showing the effectiveness of these metrics on this set of targets and on all targets in general. One of the underlying goals of discovering metrics is to reliably compute information about images (such as probability of detection (Pd) of an object) and effectively assist the soldier in his assessment of his own Pd. If a computer could accurately predict Pd, it would be able to expedite the process of concealing ground vehicles within enemy territory and supporting tactical planning. This also helps cut down development time on new camouflage treatments and concealment methods. All of this is critical to the survivability of the soldier and ground vehicles. The metrics in this paper, called Target Structure Similarity Metrics (TSSM), are derived from the popular Structural Similarity Metric. These metrics take into account certain hypothesized characteristics of the human vision system (HVS), such as sensitivity to edges and sensitivity to areas of high contrast. TSSM use these qualities to measure image quality by comparing a non-distorted reference image against a distorted image. The metric is then a measurement of how closely specific qualities of the distorted image resemble those of the reference image. The TSSM clutter metric is based on the signal processing features of human vision aided by computer comparison of the images.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 14, 2006
Accession Number
ADA454643

Entities

People

  • Euijung Sohn
  • Ivan Wong
  • Kimberly Lane
  • Thomas Meitzler

Organizations

  • Tank-automotive and Armaments Command

Tags

Communities of Interest

  • Air Platforms
  • Ground and Sea Platforms
  • Human Systems

DTIC Thesaurus Topics

  • Abstracts
  • Air Force
  • Air Force Facilities
  • Combat Vehicles
  • Computers
  • Contrast
  • Detection
  • Engineering
  • Experimental Data
  • Ground Vehicles
  • High Resolution
  • Military Vehicles
  • Perception
  • Probability
  • Signal Processing
  • Vehicles
  • Visual Perception

Fields of Study

  • Computer science
  • Engineering

Readers

  • Computer Vision.
  • Sensor Fusion and Tracking Systems.
  • Software Engineering.