Abstract
The computer image processing is useful for identifying each particle from a picture of overlapping particles like particles of fine powder or droplets of atomized liquid. For this purpose we propose a new approach in which a binary image of overlapping particles is approximated in one of overlapping circles. This approach selects the best among a number of alternative patterns of overlapping circles, each of which is produced based on a combination of numbers of circles and their intersections. The validity and performance of this approach are examined through both a real picture and some computer simulation.
