The method on stacked particle image segmentation and particle size measurement

Abstract

In the measurement of overlapping particle size, it is necessary to perform image processing on the stacked particle image to obtain accurate measurement results. In this paper morphological filtering is used to remove the isolated small area and fill the holes. The distance image combined with h-minima transform is used to get the seed points. Then the seed points and background are marked on the distance image. Finally, the distance image is segmented by watershed. Due to the partial missing after segmentation of conglutinated particles, according to the prior knowledge that the shape of particles is similar to ellipse, this paper reconstructs the contour of the incomplete particles by ellipse fitting technology. Finally, the measurement algorithm of particle shape and particle size characteristics is determined. Two groups experiments are carried out for particle size measurement, and the error of particle size measurement is analyzed. It is proved that the measurement is accurate.