Premium
A comparison of some methods for measuring the size distribution of two‐dimensionally clustered latex particles in TEM images using mathematical morphology
Author(s) -
Froehling P. E.
Publication year - 1991
Publication title -
journal of microscopy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.569
H-Index - 111
eISSN - 1365-2818
pISSN - 0022-2720
DOI - 10.1111/j.1365-2818.1991.tb03194.x
Subject(s) - pixel , mathematical morphology , materials science , image processing , micrograph , particle (ecology) , projection (relational algebra) , biological system , electron micrographs , binary image , optics , image (mathematics) , computer science , artificial intelligence , physics , algorithm , scanning electron microscope , electron microscope , composite material , geology , biology , oceanography
SUMMARY The particle size distribution of polymer latexes of, for example, polybutadiene can be measured on transmission electron micrographs using image analysis methods. By taking suitable measures in preparing the sample it can be ensured that the image only consists of projections of two‐dimensional clusters of adhering latex spheres. For the deagglomeration of these clusters a number of methods based on mathematical morphology are discussed. Speed and accuracy of the methods were tested on a personal‐computer‐based image analysis system and compared to manual measurements. Repeated openings or erosions applied to a binary image are relatively fast global methods with a low accuracy due to the fact that particle diameters are measured in whole numbers of pixels; only the horizontal diameters of rectangular pixels are measured. With an iterative method the individual components of the clusters are found by repeated conditional thickening of the erosion kernels. The method is accurate but very slow due to the large number of processing steps. The watershed method as applied to the original grey‐scale transmission electron micrograph makes use of the sharp local maxima which arise at the interface of the projection of two adhering particles. This method is as accurate as the iterative method, but about eight times as fast.