A Python-based algorithm for automated grain size analysis in ZnO-based varistor ceramics

Abstract Standard methods for grain size determination in microstructures, including those defined in ASTM E112, are largely manual and operator-dependent. In this study, commonly used edge detection and segmentation algorithms, such as the Canny detector and the watershed method, were evaluated with respect to automated grain analysis. Based on this assessment, a Python-based algorithm was developed to enable fully automated detection of grain boundaries in confocal microscopy images of ZnO-based varistor ceramics. The proposed workflow combines image preprocessing and marker-controlled watershed segmentation with automated area-based measurements of labeled grains. Each grain was treated as an individual object, its area was measured in pixels and converted to physical units, and the grain size was expressed as the equivalent circular diameter. The algorithm demonstrates reliable performance, accurately detecting intergranular boundaries and computing grain size in microstructures with variable contrast and artifacts. Owing to its modular design, the method can be readily extended for additional applications. The algorithm is freely available for academic and non-commercial use, and users are requested to cite this work when applying the method or code.