Point cloud editing using Statistic x84 to remove outliers

[Rebecca_H]

Hi everyone,

I'm relatively new to Zephyr and I'm looking for a way to remove outliers or 'noise' from my point cloud. For context, this is a photogrammetric survey in a moisture-rich environment that has resulted in 'fuzz' around the structure that I'd like to remove. It's therefore not selectable manually, or by plane.

All guidance/tutorials suggest using the Selection by points tool and applying the Statistic x84 filter, but I can't find any information on the parameters (scale, neighbours). At the moment, I'm adjusting these numbers through trial and error, without really understanding how they work. As a result, as well as selecting outliers, I'm also selecting key architectural details that I'd prefer not to delete!

Any help/suggestions on this greatly appreciated!

[Elliot]

Hi Rebecca,
here a brief description:

Select by Points - Method: Statistics X84:

The algorithm selects those points whose density doesn't meet the X84 statistic criteria.
It calculates which are the nearest points (neighbours) or not useful ones (outliers) to remove those too far from the correct solution. For each individual point in the cloud, the algorithm finds the nearest "neighbor" point (defined by the neighbors parameter) and calculates its distance. The set of calculated distances is processed through the statistical rule X84. The filter considers points that fall within a certain deviation as valid and classifies those that fall within the "tails" of the distribution as outliers (to be removed).

The parameters:

- Neighbors: Defines which neighbor points to consider for distance calculation, influencing sensitivity to local density.

- Scale: it's the scale, or "tolerance" of the X84 filter.

• High Scale: The filter is more tolerant. Fewer points are removed.
• Low Scale: The filter is more aggressive. More points are considered outliers and eliminated.