Automatic classification of female body shape using 3D anthropometric scan data

Understanding individual differences in body shape is essential for improving apparel fit, yet conventional sizing systems often fail to reflect the full spectrum of morphological diversity. This study presents a data-driven framework for classifying female body shapes using three-dimensional (3D) body scan data. A dataset comprising 1,019 Korean women was processed to extract key anthropometric ratios, including chest-to-hip, hip-to-waist, and depth-based indices. After filtering out incomplete and outlier data, 815 valid samples were analyzed using analysis of variance (ANOVA) and decision tree algorithms to derive body shape classification rules. The resulting taxonomy includes nine distinct shape types, which were further consolidated into three macro body shape groups based on morphological dominance: Top Hourglass, Regular Hourglass, and Bottom Hourglass. Canonical discriminant analysis achieved an overall classification accuracy of 93.6%, suggesting that the classification model is robust. This study provides a statistically grounded and reproducible methodology for body shape classification, offering a practical foundation for apparel pattern customization and mass-customized garment production.