Output details
11 - Computer Science and Informatics
Robert Gordon University
Automatic 3D Facial Model and Texture Reconstruction from Range Scans
We present a new template-fitting approach to deform 3D facial images and clone facial expressions, based on physical principles of elastic energy. Our approach is applied successfully to deform a generic 3D facial model to range scans of human faces with different poses and facial expression. This technique and unlike other common approaches is fully automated, and requires no user intervention. Testing was done using e.g Face Recognition Grand Challenge Database 2.0 (one of the most comprehensive 3D Face database available) and shows a clear and significant improvement in speed and accuracy, which is of interest to the research community.