AI-Based Automatic Laminate Design System Based on 3D Facial-Oral Registration Data and Aesthetic Constraints

Dental laminate restoration requires strict thickness control, enamel preservation, and facial aesthetic harmony, yet conventional CAD workflows remain operator- dependent and time-consuming. This study proposes an AI-based automatic laminate design system that integrates three-dimensional facial and intraoral scan data through facial-oral registration and generates laminate designs under explicit aesthetic and physical constraints. The framework includes facial-oral landmark detection, PCA-based initial alignment, point-to-plane ICP refinement, automatic extraction of eight key aesthetic parameters, a constraint-aware autoregressive transformer decoder for laminate mesh generation, NSGA-II-based multi- objective optimization, and FACS-based dynamic facial expression simulation. Results showed a reduction in design time from over 30 min to 3-5 min, a registration error of 0.18±0.06 mm, 98% fulfillment of thickness constraints, an aesthetic score of 8.7/10, and a 97% occlusal interference-free rate. The system improves efficiency and consistency while supporting clinically meaningful aesthetic decision-making in aesthetic restorative dentistry.