ANR JCJC AAPG 2018
ANR-18-CE23-0007-01
December 2018 - March 2024
This project focuses on the geometry of digital surfaces, which are boundaries of voxel sets. These data mainly come from the segmentation of 3D digital images. Keeping the digital nature of the data is often an advantage. However, a drawback is its poor geometry at any resolution. The challenge is to enhance its geometry by estimating extra data for each surface element, such as a relevant normal vector. The idea is to gather the geometrical information around each surface element within a patch of adaptive size: a piece of digital plane that locally fits to the digital surface. The covering of a digital surface by maximal pieces of digital plane is however hard because of their combinatorial explosion. An opportunity to make a breakthrough in this issue is the recent development of plane-probing algorithms. Based on these algorithms, we propose a new way of analyzing digital surfaces without any parameter. We expect a positive impact in graphics and 3D image analysis.