목적: To develop an augmented reality–based osteotomy guide that enhances surgical accuracy, especially for complex or revision facial bone procedures, and reduces incorrect osteotomies and related complications. 방법:In this study, we developed and validated a comprehensive system to improve the accuracy of facial bone osteotomy through dual-sensor integration, AI-based registration, and augmented reality (AR) visualization. The system comprises dual-sensor hardware with an RGB-D camera for depth and color data acquisition, navigation software, and an AR platform with real-time tracking capabilities. The hardware components, including sensors and navigation systems, were mounted on two custom-designed carts. Calibration protocols involving standardized patterns and controlled lighting were implemented to enhance the precision of depth measurements. Five three-dimensional (3D) models of the face and facial bones were created from both patient imaging data and 3D-printed replicas, reflecting diverse anatomical variations such as male, female, obese, and slender facial structures. An AI-based automatic registration algorithm aligned virtual and physical anatomy in real-time using these models. System verification was performed using both anatomical models and clinical data. Two otolaryngologists evaluated alignment errors in augmented views. Average errors were evaluated for shape sensing error (SSE), fiducial registration error (FRE), and target registration error (TRE), all of which were found to be within 2 mm. A skin deformation probability map was devised to predict areas of potential movement during manipulation to address intraoperative anatomical shifts. Real-time synchronization between the dual-sensor system and augmented reality head-mounted display (HMD) was achieved using continuous tracking algorithms. Wireless protocols enabled real- time transmission of osteotomy guides, including instrument trajectories, to the HMD. By combining precise 3D modeling, robust calibration, AI-based alignment, and dynamic AR visualization, this methodology demonstrates the potential to streamline surgical workflows, reduce errors, and improve clinical outcomes. 결과:TThe system demonstrated consistent accuracy in facial bone osteotomy procedures, as validated through comprehensive validation. The system achieved an average shape sensing error (SSE) of 2.14 ± 0.40 mm, fiducial registration error (FRE) of 2.10 ± 0.45 mm, and target registration error (TRE) of 1.81 ± 0.33 mm. These results confirm the system's reliability and precision, indicating its potential to enhance surgical guidance, reduce errors, and improve patient outcomes in complex osteotomy procedures. 결론:This study successfully developed and clinically validated a dual- sensor registration system, overcoming limitations in accuracy and shadowing effects of conventional navigation systems. Based on these results, the methodology has the potential for expansion to various surgical fields, including neurosurgery, orthopedic spinal surgery, and oral and maxillofacial surgery, enhancing precision and surgical outcomes.