TerraMobilita project — "3D mapping of roads and urban public space, accessibility and soft mobility " — responds to requests of knowledge about urban roads for soft mobility applications.

The project is built around two main areas: i) To develop new methods and tools to create and update urban maps using laser scanning and digital imagery; ii) To develop innovative applications for soft mobility planning.

The Research & Development assembly is coordinated by ARMINES Laboratories. The project was certified by the clusters Cap Digital and Advancity and was selected to the call for projects FUI 11 in 2011. The project brings together eight partners: 1 SMEs: STAR-APIC (project-chef), 3 big companies: THALES TRAINING SYSTEM, MENSI-TRIMBLE, DRYADE (Veolia), 3 public institutions: IGN, ARMINES/MINES ParisTech (laboratories CAOR-CMM-CAS), the Sciences Po Foundation, and 1 association: the CEREMH.

CMM-MINES ParisTech contribution

The CMM - MINES ParisTech contribution in TerraMobilita project is part of Andrés Serna PhD thesis, supervised by Beatriz Marcotegui. It is being developed at the Center for mathematical morphology (CMM) at MINES parisTech, France.

3D Semantic Analysis

Thanks to new 3D data availability, an increasing number of geographic applications such as Google Earth, Microsoft Bing Maps, OpenStreetMaps and Geoportail is flourishing nowadays. Some of these applications do not only require to look realistic, but have also to be faithful to reality. Automatic and semi-automatic methods for semantic analysis of these data are required in order to build accurate large scale 3D city models. We work in developing automatic methods for detection, segmentation and classification of urban entities from 3D point clouds and images.

Object detection and Digital Terrain Model (DTM) generation

Ground segmentation is a critical step since urban objects are assumed to be located on it. Then, objects are detected as ground discontinuities. Finally, a filtering step is performed in order to reduce noise but preserving thin vertical structures.

Ohio database, (USA) Object detection

Segmentation of urban objects

Using our detection approach, it is possible to get several objects, close to each other, merged in a single connected component. Therefore, a segmentation process is required. An object is considered segmented if it is correctly isolated as a single object, i.e. connected objects are correctly separated (there is no sub-segmentation) and each individual object is entirely inside of only one connected component (there is no over-segmentation).

Object detection and segmentation. Rue Vaugirard, Paris.

Classification of urban objects

Objects are classified in several categories using a supervised approach with geometrical and contextual features. Our features have a geometrical meaning and they can be adapted to any XYZ point cloud. Thus, the classification can be easily generalized, i.e. training in a database and testing in another one. This is a significant advantage because the model learned for a database can be applied to another one, even acquired by a different acquisition system, without the tedious manual annotation.

Object classification. Rue Soufflot, Paris.

Accessibility analysis for diffirent types of mobility.

One of the aims of TerraMobilita project is planning itineraries for different types of mobility, including soft--mobility. Therefore, curb detection and characterization is a very important task because it determines the suitability of a path. For example, sidewalks without access ramps may be appropriate for rollers but not for wheelchairs. Additionally, obstacles on sidewalk represent physical barriers to free mobility. In our work, we define the accessibility according to curbs geometry and obstacles on the street (where the most critical case is the accessibility for wheelchair users).

Curb detection and accessibility analysis. Rue d'Assas, Paris.

Associated publications

  • A. Serna and B. Marcotegui. "Detection, segmentation and classification of 3D urban objects using mathematical morphology and supervised learning" ISPRS Journal of Photogrammetry and Remote Sensing, 2014, Volume 93, July 2014, pp 243-255. paper. paper (publisher website) - paper (HAL archives ouvertes)

  • A. Serna and B. Marcotegui. "Urban accessibility diagnosis from mobile laser scanning data" ISPRS Journal of Photogrammetry and Remote Sensing, Volume 84, October 2013, Pages 23-32, ISSN 0924-2716, paper. Best 2013 ISPRS paper, The U. V. Helava Award. download
  • A. Serna, B. Marcotegui, E. Decencière, T. Baldeweck, A.-M. Pena, S. Brizion. "Segmentation of elongated objects using attribute profiles and area stability: application to melanocyte segmentation" Pattern Recognition Letters. Special Issue on Advances in Mathematical Morphology. 2014. Volume 47, 1 October 2014, Pages 172-182. paper.

  • A. Serna, B. Marcotegui. "Attribute controlled reconstruction and adaptive mathematical morphology". In proceedings of ISMM2013: 11th International Symposium on Mathematical Morphology. pp. 205-216. May 27-29, 2013, Uppsala, Sweden. paper - presentation
  • A. Serna, B. Marcotegui, F. Goulette, J.E. Deschaud. "Paris-rue-Madame database: a 3D mobile laser scanner dataset for benchmarking urban detection, segmentation and classification methods", 4th International Conference on Pattern Recognition, Applications and Methods, March 2014, Angers, France. paper. slides. download database.
  • A. Serna, "Semantic analysis of 3D point clouds from urban environments: ground, facades, urban objects and accessibility". PhD thesis. MINES ParisTech. December 2014, Fontainebleau, France download


For further information about CMM - MINES ParisTech contribution in TerraMobilita project, please contact Beatriz MARCOTEGUI or Andres SERNA.