High-accuracy multi-spectral thermometry via 25-band multi-spectral filter array imaging

We present a radiant temperature measurement technique utilizing a 25-band filter array imaging system, which effectively improves the temperature measurement range and temperature measurement accuracy. The CMOS sensor of the multi-spectral camera is uniformly photolithographed with a 25-band filter array, allowing for the simultaneous acquisition of spectral information and radiation intensity from the temperature source in a single image. A calibration experiment was conducted using a standard blackbody source to determine the relationship between the instrumental response values and the theoretical radiation intensity across each wave band of the multi-spectral camera. Additionally, a two-dimensional temperature field distribution was obtained using the dual annealing algorithm. Temperature inversion and error analysis were performed on multi-spectral images captured at four different temperatures, successfully yielding the two-dimensional temperature distribution of a butane flame with a maximum temperature measurement error of 0.65%, and the temperature measurement system extended uncertainty is 2.98%. The main factors affecting the temperature measurement accuracy include the accuracy of the blackbody furnace, the response noise of the multi-spectral camera, the accuracy of the filter, and the accuracy of the dual annealing algorithm. In the future, the temperature measurement accuracy of the measurement system can be improved by calibrating the high-precision blackbody, improving the signal-to-noise ratio, improving the optical accuracy of the filter, and optimizing the inversion accuracy of the algorithm. This method enables non-contact, high-precision temperature distribution measurements, offering a novel approach for accurate temperature measurement in the future.