Open tangeroo opened 4 years ago
That's not an easy question. If your data comes from range-scans, you can try constructing something like the visual hull or the depth hull. (E.g. compute a voxel grid with voxels measuring occupancy. For each voxel and each scan, you can mark the voxel as unoccupied if either it is between the camera's position and the corresponding point on the range scan, or if the voxel doesn't project onto the range-scan at all.)
Otherwise, the situation is harder. If you have a prior for envelope that encloses the shape then you can use that...
my original intention is taking some edges as constraints for building boundaries, so I only have some .ply format edges as priors. In this project, can any advise make this work. Thank you.
One thing you can try doing (though this is not the same as adding an envelope, and may be challenging to implement) is to try to introduce points on the edges as value screening constraints where the implicit function should have value zero. (I'm assuming that edges are known to be outside the shape.)
Thank you, I get it!
Hi, I'm very new in this field and I wanna input envelope files when processing PoissonRecon. My points were generated by dense matching but I don't know how to produce an envelope. Any advice I would appreciate it. Thank you.