3D facial pose estimation by image retrieval
Nemanja B. Grujic, Slobodan Ilić, Vincent Lepetit, Pascal Fua · 2008
We propose an approach to 3D facial pose estimation that, unlike most state-of-the-art techniques, can handle arbitrary poses, including extreme out-of-plane rotations, background clutter and facial expressions. It relies on a large database of registered face images of different people viewed from several perspectives. We use a powerful image retrieval technique to match the input image against database ones, which returns the most similar 3D pose. This 3D pose can then be refined using matches between input image and database images. We will show qualitative and quantitative results using images of people who do not appear in image database. 1.