You are here
Semantic segmentation of 3D data
The use of 3D data (point clouds, polygonal models, etc.) is nowadays widely diffused for various applications and in different fields, from the documentation of cultural heritage to autonomous driving, from urban planning to semantic 3D modeling. Nevertheless, to provide useful 3D data, it is necessary to associate some semantic information that can help operators to better understand and interpret the data. The research activity aims at carrying out an optimal, repeatable and reliable segmentation and classification procedure to manage various types of 3D survey data and associate them with heterogeneous information and attributes to characterize and describe the surveyed scene. The developed methods are based on supervised / data-driven machine learning methods and were tested on various scenarios, from small heritage objects to architectures or part of cities, providing for 3D classified (labbeled) data.
- E. Ozdemir, F. Remondino, 2018: Segmentation of 3D photogrammetric point cloud for 3D building modeling. ISPRS International archives of photogrammetry, remote sensing and spatial information sciences, Vol. XLII-4, in press
- E. Grilli, D. Dininno, G. Petrucci, F. Remondino, 2018: From 2D to 3D supervised segmentation and classification for cultural heritage applications. Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., Vol. XLII-2, pp. 399-406
- Grilli, E., Menna, F., Remondino, F., 2017: A review of point clouds segmentation and classification algorithms. ISPRS Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., Vol. XLII-2-W3, pp. 339-344