Feasibility of Using Linear Structural Equation Models of Target Acquisition Performance.

Abstract

This report documents the results of a preliminary effort to evolve models of target acquisition performance based on scene and target characteristics. The input data were subjective ratings of scene and target characteristics obtained from trained observers on imagery for which target acquisition performance data was available from previous experiments. A statistical technique heretofore not used in target acquisition modeling, linear structural modeling, was used to deal with the statistical problems caused by intercorrelations among predictor variables. The thrust of the modeling technique is to attempt to discover causal relations in the set of correlations among the ratings of scene and target characteristics and acquisition performance data. Two preliminary models are presented, one based only on ratings of scene characteristics and the other based only on ratings of target characteristics. Although the models were not developed far enough to compute target acquisition performance per se, each model adequately reproduces the relevant set of correlations. The statistic used to evaluate the adequacy of the models was an incremental fit index, such that values of this statistic close to one indicate good models. The value of the index for the scene model was 0.955 and for the target model, 0.911. These values suggest that although perfect fits have not been obtained, the present models are promising and deserve further development. (Author)

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 1980
Accession Number
ADA084618

Entities

People

  • A. K. Agin
  • G. E. Corrick
  • L. A. Olzak

Organizations

  • Hughes Aircraft Company

Tags

DTIC Thesaurus Topics

  • Acquisition
  • Data Acquisition
  • Equations
  • Observers
  • Target Acquisition

Readers

  • Computer Vision.
  • Regression Analysis.
  • Theoretical Analysis.