In this project, we are combining deformable models and Markov random fields using a graphical model framework for better image segmentation. The integrated framework takes advantages of both models and generate better segmentation results in many cases.
The tightly coupled model:
The exact (yet intractable) inference
The variational inference (to decouple the original intractable model inference)
The extended MRF model (solved by the belief propagation algorithm)
The probabilistic deformable model
The optimal variational parameters
The whole segmentation algorithm is an EM algorithm solving the above equations iteratively:
Results
More results in our paper:
Rui Huang, Vladimir Pavlovic, and Dimitris N. Metaxas, "A Graphical Model Framework for Coupling MRFs and Deformable Models" in Proceedings of CVPR, Vol. 2, 739-746, 2004.
A "more" tightly-coupled model (belief propagation inference can be performed in the whole model instead of using the variational inference):
Rui Huang, Vladimir Pavlovic, and Dimitris N. Metaxas, "Hybrid Framework for Image Segmentation Using Probabilistic Integration of Heterogeneous Constraints," in Proceedings of CVBIA (LNCS 3765), 82-92, 2005.
An extension to 3D segmentation:
Rui Huang, Vladimir Pavlovic, and Dimitris N. Metaxas, "A Tightly Coupled Region-Shape Framework for 3D Medical Image Segmentation" in Proceedings of ISBI, 2006