Postdoctoral position:Roadway and sidewalk surface reconstruction from 3D laser point clouds.
The position will be for 18 months, and will start from 1st of march 2012. Research will be held at the MATIS laboratory of the Institut Géographique National in Saint-Mandé, Paris, France (94).
Deadline: 31 December2011
More information:
Context
This PhD will take place within the scope of the large TerraMobilita project (a follower of the TerraNumerica project) of the CapDigital cluster (Business Cluster For Digital Content and Services of the Paris Region) which aims at building production lines to generate automatically and semi-automatically 3D models of roadways and sidewalks for mobility analysis purposes.
Subject
The objective of the post doctoral position is to reconstruct fine Digital Terrain Models (DTM) of roadways and sidewalks in dense urban areas from laser point clouds acquired by a mobil emapping system (c.f. Figure 1 in the pdf file). The reconstructed models must be precise and reliable enough in order to identify the practicable trajectories for any kind of mobility such as wheelchair, bicycle, reduced mobility person's vehicles etc.
There are two main challenges in generating such DTMs. The first is to detect and classify the 3D points corresponding to mobile objects such as pedestrians and cars and also static obstacles such as vertical posts, telephone cabins and puddles disturbing the mobility permanently. It leads to an obstacle map. The second challenge is in reconstructing the dual surface namely the free surface. This step has to cope with two problems: occlusions and the residuals of obstacle objects. The process must intelligently propagate the surface inside the occlusion areas and be robust enough to deal with small obstacle residuals.
References
Frank Moosmann, Oliver Pink, Christoph Stiller. Segmentation of 3D Lidar Data in non-flat Urban Environments using a Local Convexity Criterion. In IEEE Intelligent Vehicles Symposium, pp. 215-220, Xi'an, China, June 2009.
Michael Himmelsbach, Felix von Hundelshausen, Hans-Joachim Wuensche. Fast Segmentation of 3D Point Clouds for Ground Vehicles. In IEEE Intelligent Vehicles Symposium,pp 560-565, San Diego, CA, USA, June 2010.
N. Champion, D. Boldo. A Robust Algorithm for estimating Digital Terrain Models from Digital Surface Models in Dense Urban Areas. International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. 36 (Part 3) , Bonn, Germany, 2006.
MATIS laboratory
The MATIS laboratory of the Institut Géographique National (IGN), which is the French national mapping agency, is one of the leading laboratories in photogrammetric computer vision, image analysis and remote sensing applied to geospatial imagery and ground based imagery (e.g., provided by mobile mapping systems). It is composed of 27 researchers, including 16 permanent researchers. The MATIS laboratory has been involved in 3D data collection for 3D city modelling for twenty years, and makes use of several distinct that have been developed during this period.
Profile
- The candidate should have a PhD degree in photogrammetry, image processing or computer vision, with knowledge and interest in pattern recognition.
- Good spoken and written English. Knowledge of French would be useful.
- Good knowledge of programming language (C++) is mandatory.
- Prior knowledge and experience in the field of 3D point cloud processing will be an asset.
Location and salary
Research will be held at the MATIS laboratory of the Institut Géographique National in Saint-Mandé, Paris, France (94).
The wages will be around 2200 € per month. The position will be for 18 month, and may start on March 2012.
Contacts
Applications should be sent to both:
- Bahman Soheilian Phone: 00 33 1 43 98 84 29 E-mail: bahman.soheilian@ign.fr
- Nicolas Paparoditis Phone: 00 33 1 43 98 83 92 E-mail: nicolas.paparoditis@ign.fr
Applications consist of a cover letter describing how your research experience is relevant to the position, recommendation letters and a detailed CV.
Deadline: 31 December 2011
Keine Kommentare:
Kommentar veröffentlichen