Advanced multispectral thermometry method based on an improved light spectrum optimizer for reliable temperature and emissivity measurement

The multispectral thermometry method offers non-contact temperature and emissivity measurement with significant advantages over traditional thermometry. However, current multispectral techniques encounter challenges in processing multispectral data, especially when executing temperature and emissivity inversions without the emissivity-wavelength relationship and achieving real-time measurements from data with numerous spectral channels. These limitations hinder the practical application of multispectral measurement techniques. In this paper, we introduce an advanced multispectral thermometry method based on an improved light spectrum optimizer. This algorithm incorporates the Cauchy distribution inverse cumulative function as a mutation factor in the search phase, which helps avoid local optima and accelerates convergence. Unlike existing methods, our approach does not model the emissivity-wavelength relationship but directly processes multispectral data in accordance with Planck's law, without relying on Wien's approximation. Moreover, our method supports real-time processing, enabling inversion of both temperature and emissivity. We validate the effectiveness of our method through simulation and real experiments, demonstrating its capability to accurately determine temperature and emissivity with minimal relative error. Real multispectral data were acquired using a self-developed Fourier transform spectrometer, capturing radiation from both a standard blackbody and a remote high-temperature target. The feasibility and reliability of our method indicate a promising enhancement in multispectral thermometry applications.