|
Analyse dimages pour létude de la dynamique de cellules souches
|
Fabien Saint-Roch
Section Informatique, EPFL |
Diploma project March 2004 |
Abstract |
In this project, we present two global methods for the tracking of high-density similar objects, mainly biological. In the first section, we deal with an active contour tracking method, based on Spline-based Snakes developed in the Laboratoire d'Imagerie Biomédicale. We show that such a method allows an easy segmentation and is an efficient method for the tracking of huge objects. Nevertheless this method is not efficient for small structures. We then address a more generic method based on the construction of a trajectory tree. This method allows determining the optimal path for each image object. However this method is strictly dependant on the segmentation quality. Therefore we propose several segmenting-tools that allow adapting the method to the most number of cases. Finally we show our results and the advantages of such a method opposed to a classic search Tree. |
|
|
|