Segmentation, Graph Clustering and Visualisation Environment for 3D Images

Encadrants : 

Occurrences : 

2020

Nombre d'étudiants minimum: 

4

Nombre d'étudiants maximum: 

7

Nombre d'instances : 

1

Domaines: 

Classical image segmentation techniques such as watershed or SLIC can be improved by the generation of a graph based on the segmentation (using e.g. Region-adjacency), which can then be clustered using techniques such as spectral graph clustering. We aim to develop a graphical environment for visualisation of 3D images, and for progressive application of said techniques in a plug-in manner.

The environment will allow for loading and saving of images, segmentation (or "label") maps, graphs, and graph clusterings (or graph "labellings"). It will additionally allow for the deployment of segmentation, graph generation, and graph clustering techniques. These techniques should be implemented as plug-ins (so that future techniques can be included); all of them are already implement in e.g. Python common libraries.

 

 

Further reading: