Three-view 3D Face Reconstruction Using Facial Landmarks and Pose Estimation
Robert Hakobyan · 2023
This paper presents an innovative approach for creating a 3D model of a person's face using three images captured from different angles.The proposed method involves detecting and extracting facial landmarks from each image, as well as estimating the pose of the face in each image.By aligning the registered images and estimated poses, a 3D model of the face can be reconstructed.This model is then refined using texture information from the input images and prior knowledge of facial structure.The approach demonstrates potential applications in areas such as cyber and physical security, computer graphics, biometrics, and forensics.It is important to consider ethical concerns related to facial image usage when applying this method.