Probabilistic Morphable Models (PMMs)
PMMs are a fully open probabilistic framework for a model based image analysis using an analysis by synthesis approach. The framework splits naturally into a component for statistical object modelling and a component for fitting such a model to a novel data. The analysis of novel data is performed by fitting statistical object models to data using MCMC optimization.
The following publications form the theoretical basis of Probabilistic Morphable Models.
Gaussian Process Morphable Models [pdf]
Marcel Luethi, Thomas Gerig, Christoph Jud and Thomas Vetter
IEEE Transactions on Pattern Analysis and Machine Intelligence (Volume: 40 , Issue: 8 , Aug. 1 2018)
a preprint can be found on ArXiv: http://arxiv.org/format/1603.07254v1
Markov Chain Monte Carlo for Automated Face Image Analysis [pdf]
Sandro Schöborn, Bernhard Egger, Andreas Morel-Forster and Thomas Vetter
International Journal of Computer Vision 123(2), 160-183 , June 2017
Applications to face image analysis
For successful real-world application, carefully set up models are needed. The following papers work out individual steps of the approach in detail for the application of face-image analysis.
Occlusion-aware 3D Morphable Models and an Illumination Prior for Face Image Analysis [pdf]
Bernhard Egger, Sandro Schoenborn, Andreas Schneider, Adam Kortylewski, Andreas Morel-Forster, Clemens Blumer and Thomas Vetter
International Journal of Computer Vision (IJCV), 2018
Efficient Global Illumination for Morphable Models [pdf]
Andreas Schneider, Sandro Schonborn, Lavrenti Frobeen, Bernhard Egger, Thomas Vetter;
The IEEE International Conference on Computer Vision (ICCV), 2017, pp. 3865-3873
Morphable Face Models - An Open Framework [pdf]
Thomas Gerig, Andreas Morel-Forster, Clemens Blumer, Bernhard Egger, Marcel LÃ¼thi, Sandro SchÃ¶nborn and Thomas Vetter
13th IEEE Conference on Automatic Face and Gesture Recognition (FG 2018). pp. 75-82.