High-Precision Center Estimation of Point Source Infrared Targets

Abstract

Several different methods of estimating the subpixel center of a point source in infrared imagery are explored. These include optimized Gaussian fitting, analytic Gaussian and paraboloid global optimum solutions, and weighted-centroid approaches. Each of these methods is applied to a variety of randomly generated point source test images through Monte Carlo simulation. Several factors are incorporated, including Gaussian noise, image saturation, and orientation angle. Using the standard deviation of the errors from the Monte Carlo runs as a metric, each of the subpixel estimation algorithms is compared. Overall, the optimized Gaussian-fitting algorithm produces the best results on clean imagery, but the weighted-centroid method is most accurate for noisy, saturated images.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 2022
Accession Number
AD1189390

Entities

People

  • Ryan Decker
  • Steven Manole

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Coefficients
  • Computations
  • Computer Vision
  • Convergence
  • Equations
  • Errors
  • Estimators
  • Filters
  • Gaussian Distributions
  • Gaussian Noise
  • Image Processing
  • Mathematics
  • Monte Carlo Method
  • Noise
  • Normal Distribution
  • Precision
  • Saturation
  • Simulations
  • Standards
  • Statistical Algorithms
  • Two Dimensional

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Geodesy