In this paper is described an automatic system for measuring the concentration of cloud condensation nuclei by computer vision, the procedures and infrastructure of the laboratory for evaluation and, finally, it is compared its measurements with other equipment taken as a reference. The measurement system developed uses for such, digital image processing techniques. The process consists of capturing a sample of atmospheric air inside a static thermal diffusion chamber supersaturated with water vapor. In this condition, the molecules of water vapor condense on the aerosol captured in the air, producing water droplets. These droplets fall by gravity and cross a laser light beam, which defines a sampling volume, making them visible. Series of images of this process is digitized and processed for identification and isolation of the droplets present in the sample volume. Then, these droplets are automatically counted and their concentration is calculated. The threshold, distance and watershed transformations techniques are used. An efficient methodology for determination of sample volume is also presented. The results of experiments performed under controlled conditions indicate that the followed procedures are appropriate in determining the concentration of cloud condensation nuclei and that the equipment developed is effective in high concentrations where other equivalent instruments are no longer reliable due to the effect of overlap of droplets in the images analyzed.
digital image processing; watershed transformation; distance transform; atmospheric aerosol; meteorological instrumentation; cloud condensation nuclei
