Instrumentation and procedures for validation of synthetic infrared image generation models

摘要

Abstract: Synthetic infrared image generation models are becoming more complex with the incorporation of radiation propagation, thermodynamic, environmental, energy matter interaction, and sensor models linked through ray tracers into CAD models of scenes. As these models evolve, it is becoming increasingly necessary and difficult to design validation experiments to determine how well the models work and where the limitations are. This paper describes an experimental approach to validation of the radiometric integrity of an end-to- end thermal infrared SIG model. The approach attempts to break down the overall SIG model into a set of submodels with measurable input and output parameters. A scene is then instrumented and imaged in a time lapse fashion over an extended period (e.g., 48 hours). This scene is also synthetically produced so that the actual and synthetic scenes can be compared. The experimental approach includes acquisition of meteorological data (air temperature, relative humidity, wind speed, cloud cover, precipitation type and rate, total isolation, diffuse isolation), object data (emissivity, absorptivity, thermal conductivity, specific heat, temperature), atmospheric data (transmission, path radiance) and image data (calibrated longwave infrared and midwave infrared images, as well as visible images). Error propagation models are used in conjunction with the experimental data to determine the source and relative importance of errors in the modeling process.!13