Homework 03: Ground filtering

Deadline is 2020-01-20 13:00.

Late submission? 10% will be removed for each day that you are late.

20% of final marks

You’re allowed for this assignment to work in a group of 2 (and thus submit only one solution for both of you). If you prefer to work alone it’s also fine.

Overview

In this assignment you need to:

  1. automatically classify part of the AHN3 point cloud into a ground and a non-ground part using ground filtering (as described in Section 12.4 of the book),
  2. use nth point cloud thinning to speed up processing times (Section 12.2),
  3. make rasters from the ground points using IDW and TIN interpolation (Sections 4.3.2 and 4.3.3).
  4. compare your ground rasters with the official AHN3 DTM rasters.

Data and code skeleton

hw03.zip

The zip file for this assignment contains:

  1. geo1015_hw03.py is the main(). You are not allowed to modify this file!
  2. my_code_hw03.py is where all your code should go. You have to complete the filter_ground() function.
  3. params.json file that contains the parameters to be used by your program (see next section)
  4. example_raster_output.asc example raster output file
  5. ahn3/rural.las input point cloud with just hills and trees. (The original AHN3 file contains the classification but we removed it from the file.)
  6. ahn3/urban.las input point cloud of an urban area. (The original AHN3 file contains the classification but we removed it from the file.)

For the last step of the assignment, ie comparing your results to the official AHN3 DTM raster, you will need to download the AHN3 0.5 meter DTM raster from the AHN3 download page. The tiles that you need are 40bn2 and 37fz1 in the GeoTIFF format.

Parameters

The params.json file in the zip contains all the parameters that are user configurable. This way there is no need to hardcode any parameters in your python files. The geo1015_hw03.py script reads this file and passes its contents as a dictionary to the filter_ground() function that you need to implement.

It contains the following parameters:

All parameters have reasonable default values. You are encouraged to experiment with different values during development. All distances are in data units (meters in our case).

Allowed python libraries

Apart from the standard Python libraries you may only use:

File formats and visualisation

For point clouds you will use the industry standard LAS file format. Use the laspy library to read and write. You can visualise a LAS file with CloudCompare or Displaz, both open-source software.

For visualising the TIN that is needed in the ground filtering algorithm, you can use the OBJ format. The startin library offers a write_obj() method that makes this very easy. Visualise with CloudCompare, MeshLab or Displaz.

The raster files that you need to create should be in the Esri ASCII raster format (.asc), see its full specifications. It’s a simple ASCII/text file that can be easily created with Python, and QGIS reads it natively. See the included example_raster_output.asc for an example.

Good to know

The report to submit

You need to submit a very simple report explaining briefly how you implemented the assignment, and how your best results compare to the official AHN3 DTM raster.

We expect maximum 3 pages for this.

Marking

  Marks
followed all rules (names, functions, etc.) and runs w/o modifications 1
ground filtering 3
TIN interpolation 1
IDW interpolation 1
No border artefacts 1
report quality/clarity 3

What to submit and how to submit it

You have to submit a total of 6 files (you can put them in a ZIP file):

  1. the Python file my_code_hw03.py, where your name, and that of your colleague, is clearly identified at the top of the file.
  2. report in PDF format (please no Word file)
  3. urban-idw.asc, IDW interpolation output for urban.las.
  4. rural-idw.asc, IDW interpolation output for rural.las.
  5. urban-tin.asc, TIN interpolation output for urban.las.
  6. rural-tin.asc, TIN interpolation output for rural.las.

For all the .asc files the cellsize must be 0.5m, the other parameters (search circle, power) are up to you, try to get the best results you can!

Do not submit your assignment by email, but upload the requested files to this Dropbox file request page. Make sure that you put the full name of one of the member of the team (only one is sufficient). You’ll get a confirmation email when everything has been successfully uploaded, keep it in case things go wrong.

[last updated: 2019-12-17]