Section: Highlights of the Year

Highlights of the Year

Yuliya Tarabalka has joined our team since January 2015, making our initial objective to implement a synergy between geometry and image/vision more concrete.

On robust reconstruction of complex shapes and scenes we obtained a wide range of new results. One noticeable result is an approach for reconstruction of indoor scenes, which received the U. V. Helava Award Best Paper 2014. In our quest for semantized reconstruction we contributed an approach for reconstruction of levels of details of urban scenes, in accordance to the CityGML format. This work has been published in ACM Transactions on Graphics, and presented at ACM SIGGRAPH 2015. We also contributed a robust 3D reconstruction approach for underwater scenes, the latter exhibiting many new challenges in terms of data defect such as uncertainty and unprecedented level of outliers. Finally, we contributed a STAR (state of the art) report at the EUROGRAPHICS conference on 3D reconstruction from point clouds, which is being converted into a survey for the Computer Graphics Forum journal.

Our two-year efforts on the problem of isotopic shape approximation have turned into a new publication at the premier venue in Computer Graphics: the ACM SIGGRAPH conference 2015, and a patent. We derived a novel algorithm that generates a surface triangle mesh, given an input tolerance volume guaranteed to be within the tolerance, intersection free and topologically correct. Despite being a long standing problem, there was still no robust and practical solution to this enduring scientific challenge. This problem is both relevant to, and timely for, the increasing variety of industrial applications that involve raw geometric data.

The scientific impact of our contributions is illustrated by publications in premier journal and conference venues in our field, both in geometry processing and computer vision: ACM Transactions on Graphics and SIGGRAPH, Computer Graphics Forum, EUROGRAPHICS, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IJCV, IJRS/ISPRS. Note also that our work on underwater reconstruction has been published in the premier journal on robotic research.

Awards

Xavier Rolland-Nevière obtained the best student paper award for “track IFS” of ICASSP (IEEE International Conference on Acoustics, Speech and Signal Processing) .

Sven Oesau, Florent Lafarge and Pierre Alliez received the U.V. Helava Award Best Paper for year 2014: “Indoor Scene Reconstruction using Feature Sensitive Primitive Extraction and Graph-cut”. ISPRS Journal of Photogrammetry and Remote Sensing, 2014.

Best Paper Award:

[16]