Try the 'Tutorials for pointcloud processing in Python' now right in your browswer without installing anything on your own machine or signing up!
Click the 'launch binder'' badge below:

The use of pointclouds tends to increase over the years, as 3D acquisition systems and 3D modeling software become more widely available. Pointclouds are nowadays used in many areas, such as computer-aided design, metrology, extended reality, robotics, and autonomous driving, to name just a few.

These tutorials have been created for those wishing to learn a little bit more about the basics of pointcloud processing. Having gone through this stage during my Ph.D., I hope here to share some of what I have learned so far.

The notebooks have been designed to make pointcloud processing algorithms easier to understand, without compromising performance too much and trying to minimize the use of specialized third-party software. They require basic knowledge of Python and its main scientific libraries.

Tutorials are broken down as follows:

1. Basic operations: pointcloud attributes, transformations (translations, rotations, reflections, scaling) & subsampling.
2. Spatial indexing (50% done): distance, neighborhoods, "brute force" search, voxel grids, (octrees & kd-trees).
3. Registration: corresponding sets & non-corresponding sets (Iterative Closest Point & Principal Axis Alignment).
4. Segmentation: Region Growing, Hough Transform & RANSAC.
5. Primitive fitting: Plane, Sphere, Cylinder, Cone & Torus.

The code is in python and relies on numpy, scipy, matplotlib, and jupyterlab.

These dependencies may installed with pip with

pip install numpy scipy matplotlib jupyterlab

or via conda with

conda install numpy scipy matplotlib jupyterlab

JupyterLab may be started using the terminal or Anaconda prompt simply by typing

jupyter lab

The lists below do not pretend to be not exhaustive but may be a good starting point for those who whish to dive deeper in the topic of pointcloud processing.

Libraries:

• Open3D, an open-source library for 3D data processing (in C++ and Python, with a 3D viewer app).
• CGAL, an open-source library for efficient and reliable geometric algorithms (in C++, with Python bidings)
• PCL, a standalone, large scale, open project for 2D/3D image and pointcloud processing (in C++, with Python bidings)
• PDAL, an open-source library for translating and manipulating pointcloud data (in C++, with Python support)
• VTK, an open-source software for manipulating and displaying scientific data (in C++, with wrappers in Python, Java and Tcl)
• PyMeshLab, a Python library that interfaces to MeshLab (see below)
• CloudComPy, a Python wrapper for CloudCompare (see below)

Applications:

• CloudCompare, an open-source 3D pointcloud (and triangular mesh) processing software (with Python scripting capabilities through CloudComPy)
• MeshLab, an open-source 3D triangular meshes (and pointclouds) processing and editing software (with Python scripting capabilities through PyMeshLab)
• ParaView, an open-source visualization application (with Python scripting capabilities)
• Blender, an open-source 3D computer graphics software that has pointcloud processing capabilities (with Python scripting capabilities)

Books:

• Poux, F. (2025). 3D Data Science with Python. O'Reilly Media.
• Liu, S., Zhang, M., Kadam, P., & Kuo, C. C. J. (2021). 3D Point Cloud Analysis: Traditional, Deep Learning, and Explainable Machine Learning Methods. Springer.
• Schneider, P., & Eberly, D. H. (2002). Geometric tools for computer graphics. Elsevier.
• Goulette, F. (1999). Modélisation 3D automatique : outils de géométrie différentielle. Presses des Mines.

Videos:

• Florent Poux YouTube channel: www.youtube.com/@FlorentPoux (last accessed in August 2024)

You are mostly free to share and reuse this work as you wish. Please do not forget to cite it if you do!

An example using BibTeX:

@unpublished{gregorio2024tutorials,
     author={Grégorio, Jean-Loup},
     title={Tutorials for pointcloud processing in Python},
     year={2024},
}