Efficient, Robust and Accurate Fitting of a 3D Morphable Model教程.pdf

Efficient, Robust and Accurate Fitting of a 3D Morphable Model Sami Romdhani Thomas Vetter University of Basel, Computer Science Department Bernoullistrasse 16, CH - 4056 Basel, Switzerland sami.romdhani, thomas.vetter
@unibas.ch Abstract Robust against non-Gaussian noise. The assumption of normality of the difference between the image syn- 3D Morphable Models, as a means to generate images thesized by the model and the input image is generally of a class of objects and to analyze them, have become in- violated due to the presence of accessories or artifacts creasingly popular. The problematic part of this framework (glasses, hair, specular highlight). is the registration of the model to an image, a.k.a. the fitting. The characteristic features of a fitting algorithm are its ef- ficiency, robustness, accuracy and automation. Many accu- Accurate, as we have already pointed out, the accu- rate algorithms based on gradient descent techniques exist racy is crucial for the application which is to use the which are unfortunately short on the other features. Re- fitting results (and, generally, the level of accuracy re- cently, an efficient algorithm called Inverse Compositional quired depends thereon.) Image Alignment (ICIA) algorithm, able to fit 2D images, was introduced. In this paper, we extent this algorithm to fit 3D Morphable Models using a novel