Image and video segmentation

Semantic full Scene labelling

Our work on semantic full scene labelling aims at modelling spatial context extracted from labelled ground truth. A cascade of learning machines (similar to auto-context models) is trained to produce a segmentation map, where subsequent classifiers learn context based on outputs of previous classifiers.

Papers:

Body part segmentation / pose estimation

Papers and more details

Probabilistic graphical models

Image segmentation and restoration taking into account as much a priori information on the image as possibly available. Probabilistic graphical models (Markov random field, Bayesian networks etc.) are particularly well suited to this task and naturally combine with very efficient global optimization methods (graph cuts, belief propagation etc.).

Papers:
  • Christian Wolf Document Ink bleed-through removal with two hidden Markov random fields and a single observation field. In IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 32(3):431-447, 2010.
  • Christian Wolf and Gérald Gavin. Inference and parameter estimation on hierarchical belief networks for image segmentation. In Neurocomputing 73(4-6):563-569, 2010.
  • Christian Wolf and David Doermann Binarization of Low Quality Text using a Markov Random Field Model. in ICPR, 2002.

Adaptive image/document binarization

Our adaptive binarization method ranked #5/43 in the DIBCO 2009 competition.

Papers:

Contact

Christian Wolf
LIRIS UMR CNRS 5205
INSA-Lyon / Université de Lyon