Identification of Sample Preparation Defects in Automated Topological Characterization of Composite Materials

Abstract

Large Area Automated Microscopy, which involves the capturing and montaging of thousands of digital image frames to create one seamless picture of large-area specimens, is increasingly being used to identify and quantify meso-structural topological features in composite materials. In this work we describe the usage of a geometrical descriptor, namely the symmetry ratio (S_r), to isolate and filter out fiber fragments from optical micrographs of cross-sections of pultruded rods reinforced with continuous carbon fibers. Damaged fibers are present in such images as a result of sample preparation and can easily be mistaken for either ‘‘small’’ fibers or ‘‘misaligned’’ fibers by typical image analysis software. Elimination of such defects removes a small but potentially misleading bimodality from the size histogram of the fiber population.