View my account

outlier point cloud

Follow
Selfi Ring
  • Edited

Hi,

I want to measure objects, but when I create a point cloud from a depth frame, I get a lot of outliers (I even used different filters from the realsense API)

I create a point cloud like this, python:

points = pc.calculate (depth_frame)

v = points.get_vertices ()

Only helps the open3d library and its method remove_statistical_outlier(nb_neighbors = 60, std_ratio = .1).

How to be?

0

Comments

4 comments

Date Votes
  • Jesus Garcia

    Hello Selfi Ring,

    The Box Dimensioner Multicam python sample may be helpful. In particular, take a look at the function "convert_depth_frame_to_pointcloud" in helperfunctions.py.

    Sincerely,
    Jesus G.
    Intel Customer Support

    0
    Comment actions Permalink
  • Selfi Ring
    • Edited

    Hello Jesus Garcia, 

    I used the function, but the cloud turned out to be stretched, although if you look from the angle of the camera, everything is fine.

    Also launched Box Dimensioner Multicam, doesn't work as I would like.

    I looked at what happens to the point cloud - the rigid transform is not correctly applied to the translation in world-coordinates.

    And the cloud resulting from pc.calculate (depth_frame) translates well. I took the transformation matrix from the Box Dimensioner Multicam

    0
    Comment actions Permalink
  • Jesus Garcia

    Hello Selfi Ring,

    Give us more details on these outliers and how they affect your measurements. How are you performing the measurements?

    The lighting conditions have a big effect on the depth performance of the L515.

    Sincerely,
    Jesus G.
    Intel Customer Support

    0
    Comment actions Permalink
  • Jesus Garcia

    Hello Selfi Ring,

    Do you still need help with this issue?

    Sincerely,
    Jesus G.
    Intel Customer Support

    0
    Comment actions Permalink

Please sign in to leave a comment.

Didn't find what you were looking for?

New post