How to extract the edge correctly? about edge_extraction HOT 2 CLOSED

k_n stands for the number of neighbors, so you should define this parameter based on the density of your point cloud. If your point cloud is very dense, then your value should be larger, otherwise if the points in your point cloud are sparse, then you should consider a lower value for k_n.
The value of thresh is based on the eigenvalues. In order to define the best threshold on your data, you can comment the line 46 of the code. Then, check the visualization of line 57. There, you can define the proper threshold based on the 'colorbar' and the visualization that you get on the edges.

It would be better to keep the thresh as a fix number and play with different k_n values first to find the proper number of neighbors, since this value can be different depends on the density of your point cloud. But, thresh is usually a same value on most of the cases.

I hope this clarifies your query, let me know if you have any further problem, and thanks for your interest in this code.

from edge_extraction.

As your suggestion, I fixed the thresh =0.03 param. I tried with different k_n from 1 to the maximum value of k_neighbors. However, it was not much difference between the trial values of k_n. I couldn't extract any information from the inner edges.
I also attached the .ply here. Can you try it?
BoxFirmaSchublade.zip

from edge_extraction.