Maximum Likelihood Detection of Electro-Optic Moving Targets

Abstract

The description of a maximum likelihood algorithm to detect moving targets in electro-optic data is presented. The algorithm is based on processing image data that are modeled as temporally stationary and spatially nonstationary Gaussian samples. Algorithm performance is evaluated in terms of the probabilities of false alarm and detection. A comparison of theoretical and experimental probability distributions for single normalized pixels shows good agreement for different clutter backgrounds (stellar, sky, mountains, and desert). Similarly, a comparison of theoretical and experimental false alarm probabilities also shows good agreement. These results validate using theoretical models to predict performance. A binary integration version of this algorithm is described and evaluated in terms of false alarm and detection probabilities. This version is suboptimum and is compared with the optimum algorithm to determine the performance loss. A processing architecture concept is also described. Electro-optic sensor, detection, infrared sensor, moving target, binary integration, velocity filter.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 16, 1992
Accession Number
ADA249442

Entities

People

  • Stephen C. Pohlig

Organizations

  • Massachusetts Institute of Technology

Tags

Communities of Interest

  • Sensors

DTIC Thesaurus Topics

  • Agreements
  • Algorithms
  • Data Sets
  • Detection
  • Detectors
  • False Alarms
  • Mountains
  • Moving Targets
  • Probability
  • Probability Distributions
  • Random Variables
  • Space Systems
  • Stationary
  • Statistics
  • Target Detection
  • Targets
  • Warning Systems

Readers

  • Image Processing and Computer Vision.
  • Sensor Fusion and Tracking Systems.
  • Statistical inference.