Design and application of an automatic extraction algorithm to evaluate diesel spray characteristics in supercritical environment

Abstract

In the present paper, a three-step algorithm is designed and applied to extract the foreground spray region and measure the spray characteristic parameters from the schlieren images of fuel spray shot in supercritical environment. The main components of algorithm are divided into three parts. The first part is foreground possibility multiple estimation (FPME) algorithm, which can eliminate the background noise and preliminarily extract the foreground spray region. This method combines the temporal and spatial information of each pixel and sets multiple thresholds to judge the possibility of each pixel as a foreground pixel. The second part is gas forced flow noise elimination (GFFNE) algorithm. GFFNE algorithm is designed according to the spatial variation of environmental density. GFFNE algorithm can eliminate the gas forced flow noise caused by spray compression, and further refine the foreground spray region. The third part is erroneous foreground region elimination (EFRE) algorithm. EFRE algorithm can eliminate erroneous foreground pixels that have been transformed into background. The experimental results indicate that satisfactory agreements between the foreground spray region calculated by the three-step algorithm and the real spray region can be achieved, and the macroscopical spray characteristic parameters are entirely consistent with the real values.