persistent structures identification in 2D and 3D
DisPerSE stands for "Discrete Persistent Structures Extractor" and it is an open source software for the identification of persistent topological features such as peaks, voids, walls and in particular filamentary structures within noisy sampled distributions in 2D, 3D. In DisPerSE, structure identification is achieved through the computation of the discrete Morse-Smale complex it can deal directly with noisy datasets via the concept of persistence (a measure of the robustness of topological features).
Although it was initially developed with cosmology in mind (for the study of the properties of filamentary structures in the so called comic web of galaxy distribution over large scales in the Universe), the present version is quite versatile and should be useful for any application where a robust structure identification is required, such as for segmentation or for studying the topology of sampled functions (like computing persistent Betti numbers for instance). Currently, it can be applied indifferently to many kinds of cell complex (such as structured and unstructured grids, 2D manifolds embedded within a 3D space, discrete point samples using delaunay tesselation, Healpix tesselations of the sphere, ...).
You can find more details about the code implementation and applications in these two articles : Theory and an Application .
The code can be downloaded here.
persistent structures identification in 2D and 3D