Open jdily opened 4 years ago
Thanks for the suggestion! Let me try to update this feature in one week.
Thanks for the help :) I think the result shown in the paper looks really promising, and it would be really helpful if we can run that reconstruction and do some comparison :)
A fundamental way is to change several things in the code including voxel-point intersection and nearest neighbor selection, which could take a while. However, there is an easy way to work around with the current code, which is what I did in my experiment:
First, I use CGAL's jet_estimate_normals to compute point normals. Then, for each point I create a very small triangle (consistent with the estimated normal), setting the point as the center. And now we convert N points to a triangle soup with N triangles and we can run our algorithm.
the workaround seems nice, I will give it a try! Thx for the update will let you know if I have more questions :)
Hey @hjwdzh , Thanks for sharing your great work. I am wondering if you provide the examples of reconstruction from range scan (and processed into a point cloud) as shown in Fig. 13 in your paper.
thx a lot.