Semantic graph and arc consistency in "true" three dimensional image labelling
Aline Deruyver, Yann Hodé · Proceedings - International Conference on Image Processing · 2002
This paper presents a new algorithm for arc consistency working on dataset containing over-segmented objects without previous knowledge about this over-segmentation. We introduce in the algorithm of Mohr and Henderson [1986] the notion of transitivity between regions which seem belonging to the same object. This algorithm has been tested on "true" three dimensional images. The tests made on a set of 20 nuclear resonance magnetic cerebral images show the reliability of this method to identify the three dimensional objects of the brain.