Multiwavelength thermometry without a priori emissivity information: from promise to disillusionment

Jean-Claude Krapez · 2025

Read Multiwavelength thermometry without a priori emissivity information: from promise to disillusionment by Jean-Claude Krapez (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-20853.txt.

Introduction

Jean-Claude Krapez, a solo author, gives a sober review in a 2025 paper on measuring heat from light at many colors. Heat cameras map the infrared glow from a scene. Turning glow into temperature needs one more fact per point: how well that surface emits, a value that is often unknown.

Using many colors at once was hoped to solve for both values at once. In fact the problem has endless valid answers without extra input. From the start, experts urged adding facts about emission, such as a fixed model for it. The paper traces how that advice was often set aside, and why the promise faded.

Catalog record

CREATORJean-Claude Krapez
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE122 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-20853

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Multiwavelength thermometry without a priori emissivity information: from promise to disillusionment Jean-Claude Krapez Abstract Infrared (IR) thermography provides 2D radiance maps of the IR radiation leaving the surfaces of a scene, based on preliminary calibration. Then, to convert radiance maps into temperature maps, we need to know the emissivity of each element of the scene conjugated to each of the detector elements in the camera’s focal plane. Compared to the single-band approach, multispectral thermography gives useful insight for detecting anomalies showing distinct spectral signatures, e.g. to detect gas leaks or assess air or water pollution. In addition to that, can spatially distributed multispectral information help solve the inverse temperature-emissivity separation problem? Multiwavelength thermometry (MWT) is known to be an underdetermined problem having a continuous infinity of solutions. For this reason, and right from the origin of MWT, it appeared necessary to introduce information on emissivity to assess temperature, for example, by means of an analytical model. Overlooking these recommendations, a number of papers appeared in the early 2000s exploring the idea that we could do without any a priori knowledge of emissivity. Growth became exponential from 2020 onward, with the allegedly successful application of neural networks, genetic algorithms, and other novel optimization methods. The aim of the present work is to recall the consequences of MWT as an underdetermined inverse problem, to highlight the errors made by ignoring them, and to bring us back to harsh reality: we must introduce information on emissivity in order to evaluate temperature accurately. Furthermore, this information has to be fully consistent with the true spectral emissivity. To this end, we propose a new method for optimal selection of the emissivity model. A series of blind tests was set up to benchmark different inversion algorithms. The results confirm the failure of the temperature-emissivity separation when no emissivity information is available, despite the multiwavelength approach, which in any case comes as no surprise. v1 is the manuscript of the Keynote paper presented at the conference SPIE 13470, Thermosense: Thermal Infrared Applications XLVII, Orlando USA, 14-17 April 2025. v2 is a review updated as of June 18, 2025 a]DOTA, ONERA, 13300, Salon de Provence, FRANCE keywords multiwavelength, multispectral, pyrometry, temperature, emissivity, blackbody, greybody, underdetermined, ill-posed, inversion, optimization, radiation thermometry † † authorinfo: E-mail: [email protected], Telephone: 33 4 90 17 01 17 1 INTRODUCTION The early 2000s saw the appearance of papers on multiwavelength thermometry (MWT), claiming that the separation of temperature and emissivity could be successfully achieved without introducing any assumptions about emissivity, i.e., without requiring any a priori information about the actual emissivity. For the sake of brevity, this approach, as well as the associated methods and algorithms, will be referred to as MWT-NEI (NEI for ’no emissivity information’), as opposed to MWT-WEI (WEI for ’with emissivity information’). At first, the number of publications on MWT-NEI increased slowly, then, from 2020 onward, an acceleration was observed, as shown in Fig. 1 . To date, at least 101 papers published in academic journals (i.e., excluding conference papers) have been identified. We here refer to those MWT methods that exclude the necessity of specifying an analytical model for emissivity. We also made a research on MWT methods based on emissivity models but applied blindly , we will call them MWT-EMAB, as opposed to MWT-EMAK where the emissivity model is applied knowingly . In Fig. 1 the count of papers on MWT-NEI methods is supposed to be exhaustive, while that on MWT-EMAB is not, by far. Figure 1: In red: number of papers on multiwavelength thermometry (MWT) claiming that no emissivity information (NEI) is necessary to estimate accurately the temperature, i.e. MWT-NEI methods. In magenta: number of papers on MWT with emissivity models applied blindly (EMAB) (by June 18th 2025). Consider the basic MWT configuration performed at n wavelengths \lambda_{i} , i=1,...,n , and without reflected radiation. The measured radiance L_{i} (assumed for the moment to be error free) of the radiation emitted in a given direction and at wavelength \lambda_{i} by a solid surface at temperature T is expressed by: L_{i}=\epsilon(\lambda_{i},T)B(\lambda_{i},T)\,,\;i=1,...,n (1) where \epsilon(\lambda_{i},T) is the surface emissivity at wavelength \lambda_{i} and temperature T and B(\lambda_{i},T) is the blackbody radiance at wavelength \lambda_{i} and temperature T . The blackbody radiance is given by the Planck’s law: B(\lambda,T)=\frac{C_{1}}{\lambda^{5}}\frac{1}{\exp\left(C_{2}/(\lambda T)\right)-1}, (2) or the Wien’s approximation: B_{W}(\lambda,T)=\frac{C_{1}}{\lambda^{5}}\exp\left(-\frac{C_{2}}{\lambda T}\right), (3) where C_{1} =1.191 \cdot 10 -16 W m -2 and C_{2} =1.439 \cdot 10 -2 m K. The Wien’s approximation is better than 1% provided \lambda T< 3124 µm K. It is well known that MWT is an underdetermined problem since we have n+1 unknowns (the n values of emissivity plus the temperature) but only n measurements of radiance. Basically, the problem has a multiplicity of solutions , and for this reason, additional conditions or constraints on emissivity must be introduced (see e.g. Ref. [ 1 , 2 ] and the reviews in Ref. [ 3 , 4 , 5 ] ). Coates declared that since the set of equations contains one more unknown than there are equations, it is therefore insoluble ”unless some relationship between these can be introduced” [ 1 ] or ”without at least one additional piece of information” [ 2 ] . The constraint on emissivity can take the form of an analytical model for emissivity with a reduced number of parameters as compared to the number of wavelengths. The solution can then be obtained by least-squares optimization. Alternatively, we can introduce a set of a priori values and uncertainties for emissivity and/or temperature and solve the problem in a Bayesian framework. When using an analytical model, it is also well known that it should be chosen very carefully to obtain satisfactory results, otherwise the error on temperature can be very high [ 3 , 5 ] . Realizing multispectral measurements at several temperature levels, with the aim of having more equations than unknowns, and thus possibly eliminating the ill-posedness of the problem, is just an illusion: with temperature-independent emissivity and applying the Wien approximation, it can be shown that the problem remains underdetermined [ 6 , 3 ] . With Planck’s formulation, it remains highly ill-conditioned, even in the presence of reflections [ 3 , 7 ] . Because of this burden, i.e. the requirement for a ”good” emissivity model, some have been tempted to search for methods that dispense that do without any emissivity information, i.e. MWT-NEI methods. However, from a mathematical logic point of view, this is nonsense, precisely because of the multiplicity of solutions. In fact, these methods and the papers describing them raise a series of questions: • how can they manage separating temperature and emissivity despite the fact that the MWT problem is underdetermined and therefore has a multiplicity of solutions ? • how can the published results obtained from simulated radiance data be so accurate ? • how can the published results obtained from experimental radiance data be so accurate as well ? Should we invoke the presence of informational biases during the inversion tests to explain these last two points ? Information bias is a risk encountered when people try to solve an inverse problem but know the solution in advance. The best way to avoid this risk is to

Read the complete text