0

Greeting,

I have sort of an opposite problem of point cloud registration. I am looking to detect whether some point cloud data that I have is disjoint. For example:

enter image description here

Is a single point cloud but is clearly discontinuous. My original thought into inspecting the distance matrix I think is moot since the "gap" might be roughly the same distance between several points (the corners) in the "individual" point clouds. The topology is subject to change so it's not safe to assume otherwise.

A literature search didn't turn up anything of interest to me. Any ideas?

PatrickR
  • 4,247
Sterling Butters
  • 35
  • Maybe I can squeeze each set of points into some "nearest-neighbor" bounding box of sorts and then count how many bounding boxes I have? – Sterling Butters Dec 05 '23 at 23:25
  • Check cluster dbscan, Google this with open3d and you will find opened tutorial on it. Number of clusters will answer your query – saurabheights Dec 06 '23 at 05:01
  • 1
    @saurabheights that’s exactly what I was looking for! Thanks! It pays to know the terminology in order to find what you are looking for – Sterling Butters Dec 06 '23 at 15:57

0 Answers0