出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：This study deals with developing an image processing algorithm that is able to recognizespherical and ellipsoidal droplets growing on pillared surfaces during heterogonous dropwisecondensation. The problem with recognizing droplets on the pillared substrates is that dropletsare very similar to the pillars or they cover several pillars at the same time, so characterizingthe pillars is very important. On the other hand the droplets are not always perfectly sphericalor they are connected and form irregular shapes, in such cases the ability to recognizing andseparating connected droplets is another challenging step. The method that is used here consistsof three main parts: pillars characterization, droplets categorizing and droplets segmentation.The result of this algorithm will be binarized images that enable to extract the informationrelated to the size and density of droplets needed for studying droplets evolution during time.Also a computer simulation method is proposed to generate ellipsoidal droplets on pillaredsubstrate. The results of this algorithm then are validated by comparing with results fromexperimental procedure.
