Biometric recognition from 3D facial images is inherently advantageous compared to 2D facial images, as it does not suffer from pose and illumination variations. However, many existing databases consist of one or more 2D facial images.
The goal of this project is to develop a method for the recosntruction of 3D facial images from two 2D facial images (front and side).
TDT4195 (Visual Computing Fundamentals)
Preferably also TDT4230 (Graphics & Visualization) or TDT4265 (Computer Vision), Machine Learning.
Supervisor: T Theoharis