Comparison of Different Ways for Image Post-Processing: Detection of Flame Fronts

A detailed understanding of the complex chemistry-turbulence interaction is gaining an increasing importance for further improvement of IC engine performance. Multidimensional optical diagnostic techniques have become a versatile tool for engine development. Sophisticated automatic data post-processing will achieve an increasing significance for efficient data reduction in such optical experiments. The focus of this paper is the detection of flame fronts using different image processing algorithms. In a further step of the data reduction, the extraction of the length of the flame front and the area of the burnt gases is presented. A strategy relying on a sensitivity analysis is discussed which allows an objective choice of parameters necessary for the application of the mathematical algorithms. As different test examples, chemiluminescence images and planar laser induced fluorescence images of formaldehyde (present in the unburnt end-gas) and OH radicals (present in the post flame region) recorded from an operating research engine are used.