Diesel spray image segmentation with a likelihood ratio test

Abstract

To characterize the macroscopic behavior of Diesel sprays and to validate and extend for current high-pressure injection systems the correlations existent in the literature, it is necessary to determine the spray geometry accurately, at least in terms of spray tip penetration and cone angle. These parameters are measured by analyzing Diesel spray images and are highly sensitive to the correct edge determination. An algorithm for segmentation of color images based on a likelihood ratio test is presented. This algorithm is compared with others available in the literature and has been validated, even for adverse experimental conditions. The experimental facilities, optical layouts, and image-processing algorithms are described.