3DTopoMap

Presentation

3DTopoMap is a software allowing of topological 3D image processing. It is a prototype implementing the results of my research works, then of the PhD thesis of Alexandre Dupas. The basic idea of these researches is that many algorithms of image processing (starting from the segmentation algorithms, but also the methods of feature computation or the modification operations) need the topology to guide and structure the processings. As topological information, we start by the simpler ones such as the adjacency between two regions, and going to much more complexes information such as homology groups.

The software

The 3DTopoMap software allows to load a 3D image and to compute the corresponding 3D topological map, a combinatorial model based on combinatorial maps that represents the topology of the regions in the image. Then it allows to:

• segment the image, using or not some topological criteria
• visualize the image, in voxels, intervoxels of 3D slices form
• simplify the image by removing the small regions
• correct the segmentation by merging or splitting manually some regions
• rework partially the segmentation result by selecting some regions and running again some split-and-merge algorithms
• compute some topological characteristics of regions (Euler characteristics, Betti numbers)
Other operations are under development.

Some objets

• Here two snapshots of the software.
• Here for the processing of a medical image. We can see the Euler characteristic of the selected volume (a torus with 125 holes to the left, and a torus with 100 holes to the right).
• Two representations of the used data structures : to the left the combinatorial map, and to the right the labeling in regions.

For more information

For more information, contact me. You can read my main publications related to this project. Lastly, the source code will be available for download here soon.

[Back]