ZED 2i Camera Point Cloud Outlier Issue

Software Python
1

Hello
I have generated point cloud data for two leaves using a ZED 2i camera, with images taken from above the leaves. However, I’ve noticed several outliers appearing beneath the leaves’ actual surfaces in the point cloud visualization. Could you provide an explanation for the presence of these outliers? Additionally, I would appreciate guidance on methods to eliminate such artifacts to improve the clarity of the point clouds both for visualization and subsequent saving of the data.

image
image824×650 55.4 KB

image
image604×636 88.2 KB

image
image872×626 77 KB

2

Hi @code_lover
the color information is affected by non-eliminable noises caused by the electrical signals and by the behaviors of the CMOS sensors.
You can reduce the number of outliers by filtering out the depth values with low confidence thanks to the runtime parameter confidence_threshold.

1 Like
3

Thanks for your guidance, by using confidence threshold I am able to improve the situation but still it is not perfect and I can observe the noise in some area. Could you please advise what should I do to remove all the noises?

4

Unfortunately, it is not possible to remove all the noises.
I recommend you design a filter that allows you to achieve your goal.
You can try to use the PCL or Open3D libraries to process the point cloud.

1 Like