Abstract—Microscopic wear particles carried by lubricating oil provide important diagnostic information about machine condition, wear mechanisms, and potential failure modes. Conventional wear-particle analysis frequently relies on professional visual interpretation, which may be subjective, labor-intensive, and challenging to replicate. This study presents an automated framework for classifying wear-particle shapes by integrating image processing, Fourier descriptor-based feature extraction, and hierarchical cluster analysis. Six common wear-particle types are considered: rubbing, severe sliding, cutting, fatigue, chunky, and spherical particles. Particle boundaries are extracted from microscopic images and represented using three Fourier descriptor formulations: Radial, Theta, and XY descriptors. These descriptors transform particle profiles into invariant coefficient features that are robust to changes in size, position, and orientation. The resulting feature vectors are evaluated using hierarchical clustering with Euclidean distance and average linkage. Comparative analysis indicates that the XY Fourier descriptor provides the most effective shape representation, producing more compact and better-separated clusters than the Radial and Theta descriptors. It then groups particles with similar morphological traits, supporting accurate and objective classification. Quantitative validation using the Silhouette Score and Davies–Bouldin Index further confirms the superior clustering performance of the XY descriptor. The proposed framework provides an objective and computationally efficient approach for wear-particle classification and supports condition-based maintenance by reducing reliance on manual expert assessment.
 
Keywords—wear particle analysis, Fourier descriptors, cluster analysis, image processing, morphological classification
 
Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).