A validation and sensitivity study was performed in which the accuracy of an infrared synthetic image generation model (DIRSIG) was examined. The majority of the error in the model is derived from two major programs: a temperature predictor and a radiative transfer model. A thermodynamic model computes kinetic temperatures of objects in a computer generated three-dimensional scene. These temperatures, atmospheric data, and sensor parameters are used as input to a ray_tracer which models the propagation of radiation in a source-target-sensor path within this scene. The output is a simulated infrared image which a sensor with the given input parameters would record. The accuracy of the model and the impact of uncertainties in individual input variables were determined by error propagation methods and comparison of the simulated imagery with actual imagery. The average theoretical error in radiance reaching the sensor was determined to be 1.58 W/m2-sr, while the measured average error in radiance for an actual predicted scene was determined to be 2.98 W/m2-sr (13.99%). Variables which had the greatest impact on the final predictive error of the model were identified and ranked accordingly. In addition, problem areas within the model were identified and suggestions for improvement were made.
Library of Congress Subject Headings
Infrared imaging--Computer simulation--Evaluation; Infrared technology--Simulation methods--Evaluation; Remote sensing--Data processing--Evaluation
Department, Program, or Center
Chester F. Carlson Center for Imaging Science (COS)
Rankin, Donna D., "Validation of DIRSIG, an infrared synthetic scene generation model" (1992). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus