|
A visualization technique was developed to accurately resolve the locations and phases of liquid and frozen water droplets naturally evolving on a subfreezing surface exposed to a humid air stream. The technique utilized the lens-like properties of unfrozen condensate droplets to focus reflected light in a manner that made it possible to reliably distinguish the unfrozen from frozen droplets. By inducing this lensing effect in all visible droplets on the surface prior to freezing, a small, high contrast region within each liquid droplet was produced. When freezing occurred, the higher opacity of the ice phase caused the frozen droplets to suddenly take on a darker appearance than the unfrozen droplets. A reflected light microscope and digital camera apparatus were used to feed data into an image analysis algorithm capable of mapping the time-evolution of the freezing water droplets on the surface. The algorithm processed raw image sequences into binary image stacks from which the light foci associated with the droplets could be reliably isolated into high contrast white pixel clusters on an otherwise black background. An edge detection method was employed within each binary image to locate the coordinates of the pixels residing along the periphery of each droplet’s focal cone. Those data were subsequently used to compute the centroidal coordinates and state of each droplet over time. Compared with more laborious, frame-by-frame analyses with various commercially available software tools, this technique correctly identified the locations and states of the droplets through the entire growth and freezing processes with accuracies in excess of 99%. This technical approach also allowed for simultaneous tracking of all droplets and freezing events present within the region of interest—typically numbering in the hundreds or thousands—over time, paving the way for interdroplet coalescence and freezing interactions to be studied in detail.
|
