Finding percentage and removing overlaping points in multiple point clouds

Hi @marcomusy,

Is it possible to use vedo to find the percentage of overlapping points and remove them from multiple loaded point clouds?

For example in the image below: I have loaded 3 different point clouds (i.e. red, orange, yellow) which they overlap to each other. Then I would like to find how much do they overlap and keep only the points where there is not overlapping area.

import numpy as np
import vedo as vd
import natsort
from wcmatch import glob

folder = ./7-scenes-redkitchen/'
pcd_files = natsort.natsorted(glob.glob(folder + '**/*.ply', flags=glob.BRACE | glob.GLOBSTAR))

tranf_files = natsort.natsorted(glob.glob(folder + '**/*.txt', flags=glob.BRACE | glob.GLOBSTAR))

pcds = []
for i, input_path in enumerate(pcd_files):
    t = np.loadtxt(tranf_files[i], skiprows=1)
   pcd = vd.Points(input_path).color(i).applyTransform(t, reset=True)
   pcds.append(pcd)

   vd.show(pcds, axes=1).close()

7-scenes-redkitchen.zip

[marcomusy]

yes, you may use mypcl.averageSize() or mypcl.diagonalSize()

[ttsesm]

Thanks Marco, I confirm that now it works as supposed.

Regarding the actual issue subject, below is a method to extract the overlapping percentage of two point clouds (considering that they are aligned) based on their points distance:

import numpy as np
import vedo as vd

def get_overlap_ratio(pcd1_, pcd2_, visualize=False):
    pcd2_.distanceTo(pcd1_).addScalarBar()

    if visualize:
        vd.show(pcd1_, pcd2_, axes=1).close()

    overlapping_points = np.where(pcd2_.pointdata['Distance'] < 0.1) # 0.1 is the distance to consider that two points are overlapping

    match_points = (pcd2_.pointdata['Distance'] < 0.1).sum() # or len(overlapping_points)

    overlap_ratio = match_points / len(pcd2_.points())

    return overlap_ratio * 100, overlapping_points

The overlapping_points contain the indexes of the points, which can be cleared out by pcd.delete() if someone wants to delete these points.