ON THIS PAGE

• 연구 배경
• Marker-Less Registraion in Mixed Reality
• NeRF (Neural Radiance Fields)
• 매핑 함수 (MLP 학습 함수)
• 손실 함수 (Loss Function)
• Process Overview
• (a) Multi-view image data acquisition
• (b) 2D Facial Landmark Extraction
• (c) Reprojected depth images and NeRF-rendered depth
• (d) 3D Facial Landmark Localization using depth images
• (e) CT Facial Landmark Extraction and 3D to 3D Registration
• Multi-View Image Data Acquisition
• 2D Facial Landmark Extraction
• 얼굴 탐지 (MTCNN)
• HRN (High-Resolution Network)
• Reprojected Depth Images and NeRF-Rendered Depth Images
• 3D Facial Landmark Localization using depth images
• CT Facial Landmark Extraction and 3D to 3D Registration
• 1. CT 데이터에서 3D 랜드마크 뽑기 (준비 단계)
• 2. SVD를 이용한 초기 정렬 (Coarse Alignment)
• 3. ICP를 이용한 미세 조정 (Fine-tuning)
• 4. 최종 결과물: 변환 행렬 (Transformation Matrix)
• Experiment
• Experiment Settings and Implementation
• 평가 지표
• Experiment Procedure
• Result
• 논문 결론