_ _ _

Reusable point cloud analytics software. Includes segmentation, registration, file format conversion. This makes use of the python bindings of the Point Cloud Library (PCL; <https://github.com/NLeSC/python-pcl>).

Copyright 2014-2015 Netherlands eScience Center. Covered by the Apache 2.0 License, see the file LICENSE.txt.

First install the required dependencies:

• Python 2.7
• NumPy
• SciPy
• virtualenv
• LibLAS
• PCL 1.7

Now set up an environment:

$ virtualenv /some/where/env --system-site-packages
$ . /some/where/env/activate

Install the python packages listed in requirements.txt using pip:

$ pip install -r requirements.txt
$ pip install -U nose  # make sure nose runs in the virtualenv
$ python setup.py install

To exit the python virtualenv run:

$ deactivate

The main functionality of PatTy Analytics is contained in the registration script. This script takes an unaligned dense point cloud and attempts to align in on to the existing drivemap. The script also attempts to find the optimal scaling, rotation and orientation of the dense point cloud, as part of the alignment process. This script can be run as follows:

$ python scripts/registration.py SOURCE.las DRIVEMAP.las FOOTPRINT.csv OUTPUT.las

where:

• SOURCE.las -- is the dense point cloud to be registered
• DRIVEMAP.las -- is the drive map where the point cloud is registered to.
• FOOTPRINT.csv -- is the footprint of the point cloud on the drive map.
• OUTPUT.las -- is the resulting registered point cloud.

additionally, an upfile.json containing the up vector (estimated from the camera position) can be provided.

$ python scripts/registration.py SOURCE.las DRIVEMAP.las FOOTPRINT.csv OUTPUT.las -u upfile.json

## Examples

The following image is a screenshot of a dense point cloud to be registered on the drive map -- this would correspond to SOURCE.las.


This screenshot shows the drive map where we want to register to -- this corresponds to DRIVEMAP.las


Finally, this screenshot shows the dense point cloud registered on the drive map. The dense point cloud has been rotated, scaled and translated to find it's best fit on the drive map -- this corresponds to OUTPUT.las.


To run unit tests, issue:

$ nosetests

Documentation can be found here