Many industrial and scientific applications require accurate measurement of radiation flux emitted by target surfaces. One of its most important applications is thermal infrared imaging for non-contact measurement and monitoring of temperature.
Emitted radiation flux is a function of the object’s temperature (via the Planck equation) and of its radiative properties (emissivity). When the emissivity of an observed object is not sufficiently known, the problem of temperature/emissivity separation (TES) is encountered. TES is perhaps the biggest challenge in infrared thermography, manifesting as an inability to decouple the effect of emissivity on the radiant flux, and in following, reliably recover temperature. Multispectral radiation thermometry (MRT) methods try to overcome this problem by assuming an emissivity model and combining measurements from several spectral bands to yield the target temperature and model parameters.
This research advances the state of the art in two avenues: by exploring better ways to operate MRT systems, and by presenting an automated methodology for their spectral design.
In terms of operation, experimental practices for removing parasitic heat flux and random electronic noise are discussed. Image fusion is shown to serve a dual purpose: both overcoming detector nonlinearities, and increasing tonal fidelity. The commonly-used NUC (nonuniformity correction) procedure is challenged and improved upon. An extended optical system calibration procedure is suggested, which captures not only the relation between temperature and detected signal, but also between temperature and the drift of the characteristic wavelength. Lastly, a mathematical model is developed for utilizing the new calibration technique in temperature recovery.
In terms of spectral design, this work presents the first use of a multi-heuristic a-posteriori optimization technique for design space exploration of MRT systems. Automated designs can be made applicable to either wider or narrower ranges of target emissivities, via user-supplied information. Superiority of automated over manual designs is demonstrated. Simulations indicate that a 4-channel thermographic system, designed and operated according to the suggested guidelines, can recover temperatures between 150-600°C within ±10°C for common aeronautical alloys (such as Inconel718, HastelloyX and Ti-6Al-4V).
The individual results of this work should all be seen as practical recommendations that can be applied to most contemporary MRT systems to improve their performance with no hardware changes (with the exception of simple optical filter adjustment).
