Brussels / 3 & 4 February 2018


Efficient and interactive 3D point cloud processing

Combining the strengths of pdal, ipyvolume and jupyter

I will demonstrate the tools we use to process large scale point cloud datasets, and our interactive workflow which enables us to quickly fine-tune custom 3D modelling algorithms.

With the advent of many extensive openly acessible point cloud datasets (like Flanders' region-wide lidar dataset), processing those point clouds becomes increasingly challenging, requiering efficient, robust algorithms. While off-the-shelf algorithm exist for common tasks like ground/non-ground segmentation, advanced 3D modelling still remains mostly in the realm of tailor-made algorithms.

Starting from established processing tools like pdal, I'll show an interactive workflow to iteratively explore and develop custom 3D modelling algorithms, through the web-based jupyter interface and in particular the ipyvolume library.

Time allowing, I will discuss some work in progress for pdal, as well as upcoming tools in the jupyter ecosystem.

This talk is going to be related to, but different from the interactive point cloud processing talk I gave at FOSS4G BE, see links and references therein.


Photo of Mathieu Carette Mathieu Carette