文摘
Noise reduction, restoration, and segmentation methods are developed for the quantitative structural analysis in threedimensions of aggregated oil-in-water emulsion systems imaged by fluorescence confocal laser scanning microscopy.Mindful of typical industrial formulations, the methods are demonstrated for concentrated (30% volume fraction) andpolydisperse emulsions. Following a regularized deconvolution step using an analytic optical transfer function andappropriate binary thresholding, novel application of the Euclidean distance map provides effective discriminationof closely clustered emulsion droplets with size variation over at least 1 order of magnitude. The a priori assumptionof spherical nonintersecting objects provides crucial information to combat the ill-posed inverse problem presentedby locating individual particles. Position coordinates and size estimates are recovered with sufficient precision topermit quantitative study of static geometrical features. In particular, aggregate morphology is characterized by a novelvoid distribution measure based on the generalized Apollonius problem. This is also compared with conventionalVoronoi/Delauney analysis.