Classification of Individual Particles Based on Computer-Controlled Scanning Electron Microscopy Data

Abstract

Computer-controlled scanning electron microscopy (CCSEM) is known as a powerful tool for measuring individual particle characterization, including various size parameters and the major elemental composition in a short analysis time. To exploit CCSEM as a receptor-modeling technique in a source apportionment study, it is important to define the membership of each particle in a well-defined particle class. Various clustering methods were examined to obtain possible members of homogeneous particle classes. An expert system was then used to build a universal classification rule based on examples of the homogeneous particle classes. These methods were extensively explored and tested using data from a study of El Paso, TX.