Reassessing nasal geometry: A PCA-based comparison of 2D dimensions and 3D point net

Poor-fitting Personal Protective Equipment (PPE) around the nose and eyes remains a persistent issue, due to the limited understanding of facial geometry diversity. This study aims to evaluate the representativeness and reliability of 2D and 3D nasal data using the Principal Component Analysis (PCA) and two-stage clustering for nose classification. A custom algorithm is developed to automate the nasal measurement from 3D facial scan data. The 3D nose point net achieved an average accuracy of 60.26 % ± 12.77 % across 8 clusters, outperforming the traditional nasal dimensions method (55.40 % ± 18.07 % across 15 clusters). These results demonstrate the superior stability and precision of the 3D approach in capturing nasal variability. This study introduces a universal classification process and key variables for nose shape analysis, offering a more holistic understanding of nasal morphology. This framework supports improved anthropometric design and has broader implications for PPE development and other facially fitted products.