An IR image processing approach for characterising combustion instability

The paper presents a first approach to the analysis of the dynamic behaviour of a pre-mixed hydrogen/air jet flame using near-IR imaging. In this kind of flames spectral lines are observed in the infrared, visible, and ultraviolet regions; emissions of visible light are however not so strong as in other wavelengths, especially in the near-infrared and infrared portions of the spectrum. Using a video camera, coupled with an optical filter, it is possible to capture images of the flame in a burner and then acquire sequences of images of the near-infrared emission during the combustion process. In this work, carried out in the frame of a collaboration between ENEL Produzione Ricerca and ISTI CNR, we propose a method suitable to process and analyse maps of near-infrared emissions in order to characterize the flame morphology and compute geometric and densitometric features useful to describe the combustion dynamics. The map of the emissions presents in some cases different unconnected sources with interesting distribution. In our analysis, we considered two different layers: the single images, as a distribution of near-IR emission, and the complete sequence of images in a dynamical evolution of the whole process.