This tool is accessible via the 'Tools > Other > Compute geometric features' menu.

Description

This tool allows to computation several geometric features on one or several point clouds (in a batch).

Geometric features that can be computed this way are:

Roughness (equivalent to the former Roughness tool of v2.9 and below - in version 2.12, an option allows to compute signed roughness values)
Curvature (equivalent to the former Curvature tool of v2.9 and below)
- Mean curvature
- Gaussian curvature
- Normal change rate
Density (equivalent to the former Density tool of v2.9 and below)
- Number of neighbors
- Surface density
- Volume density
Moment
- 1st order moment
Features
- Sum of eigenvalues
- Omnivariance
- Eigenentropy
- Anisotropy
- Planarity
- Linearity
- PCA 1
- PCA 2
- Surface variation
- Sphericity
- Verticality
- 1st eigenvalue
- 2nd eigenvalue
- 3rd eigenvalue

Notes:

most of the above features are defined in "Contour detection in unstructured 3D point clouds", Hackel et al, 2016
curvature formulas computed from a local quadric function come from: "Three-dimensional surface curvature estimation using quadric surface patches", I. Douros and B. Buxton, University College London
roughness computation algorithm is described in the old Roughness tool description page

Procedure

Select one or several point clouds then launch this tool.

A dialog will appear:

Then set the 'Local neighborhood radius' on which the selected features will be computed.

Then choose the geometric feature(s) that should be computed (several features can be selected/checked at the same time).

Notes:

currently each feature is computed independently. The extraction of the local neighborhood is sadly not leveraged for all the features...
since version 2.12, an option allows to define a 'up direction' to compute signed roughness values

On success, CC will compute the selected features one by one and create a scalar field for each.