brdr 0.17.3

BRDR - a Python library to assist in realigning geometries (OGC Simple Features) to reference borders

brdr

a Python library to assist in realigning geometries (OGC Simple Features) to reference borders

Quick links:

• Documentation & API Reference
• Installation
• Development
• Issues, questions, comments and contributions

Description

Documentation/API Reference

https://onroerenderfgoed.github.io/brdr/

Intro

brdr is a Python package that assists in aligning geometric boundaries to reference boundaries. This is an important task in geographic data management to enhance data quality.

• In the context of geographic data management, it is important to have accurate and consistent boundaries for a variety of applications such as calculating areas, analyzing spatial relationships, and visualizing and querying geographic information.
• When creating geographic data, it is often more efficient to derive boundaries from existing reference data rather than collecting new data in the field.
• brdr can be used to align boundaries from new data to reference data, ensuring that the boundaries are accurate and consistent.

Example

The figure below shows:

• the original thematic geometry (blue),
• A reference layer (yellow-black).
• The resulting geometry after alignment with brdr (green)

In the animated gif below you can see the core of 'brdr' in action:

• The visualization on the left:
  • the original thematic geometry (blue),
  • A reference layer (yellow-black).
  • The resulting geometry after alignment with brdr (green)**
• The graphic on the right:
  • X-as: Relevant distance (~distance that change is allowed), that increases
  • Y-as: Change (%) of the resulting geometry

brdr will 'detect' stable situations that result in one or more predictions.

Functionalities

brdr provides a variety of functionalities in the Aligner-class to assist in aligning boundaries, including data-loaders, processors to make predictions and export-functionalities. Besides the generic functionalities, a range of Flanders-specific functionalities are provided. The API reference and examples can be found at: Documentation & API Reference

Possible application fields

• Geodata-management:
  • Implementation of brdr in business-processes and tooling
  • Bulk geodata-alignment
  • Alignment after reprojection of data
  • Cleaning data: In a postprocessing-phase, the algorithm executes sliver-cleanup and validity-cleaning on the resulting geometries
  • Version management: visualise differences between versions of geodata
  • ...
• Data-Analysis: Investigate the pattern in deviation and change between thematic and reference boundaries
• Update-detection: Investigate the descriptive formula before and after alignment to check for (automatic) alignment of geodata
• ...

QGIS-plugin

An implementation of brdr for QGIS can be found at GitHub-brdrQ. This QGIS-plugin provides a User Interface to align thematic data to a reference layer, showing the results in the QGIS Table of Contents.

Installation

You can install the latest release of brdr from GitHub or PyPi:

pip install brdr

Basic example

from brdr.aligner import Aligner
from brdr.geometry_utils import geom_from_wkt
from brdr.loader import DictLoader

# CREATE AN ALIGNER
aligner = Aligner(
    crs="EPSG:31370",
)
# ADD A THEMATIC POLYGON TO THEMATIC DICTIONARY and LOAD into Aligner
thematic_dict = {"theme_id_1": geom_from_wkt("POLYGON ((0 0, 0 9, 5 10, 10 0, 0 0))")}
loader = DictLoader(thematic_dict)
aligner.load_thematic_data(loader)
# ADD A REFERENCE POLYGON TO REFERENCE DICTIONARY and LOAD into Aligner
reference_dict = {"ref_id_1": geom_from_wkt("POLYGON ((0 1, 0 10,8 10,10 1,0 1))")}
loader = DictLoader(reference_dict)
aligner.load_reference_data(loader)
# EXECUTE THE ALIGNMENT
relevant_distance = 1
aligner_result = aligner.process(
    relevant_distances=[relevant_distance],
)
process_results = aligner_result.get_results(aligner=aligner)
# PRINT RESULTS IN WKT
print("result: " + process_results["theme_id_1"][relevant_distance]["result"].wkt)
print(
    "added area: "
    + process_results["theme_id_1"][relevant_distance]["result_diff_plus"].wkt
)
print(
    "removed area: "
    + process_results["theme_id_1"][relevant_distance]["result_diff_min"].wkt
)

The resulting figure shows:

• the reference polygon (yellow-black)
• the original geometry (blue)
• the resulting geometry (green line)
• the added zone (green squares)
• the removed zone (red squares)

More examples can be found in Examples

Workflow

(see also Basic example)

To use brdr, follow these steps:

• Create a Aligner-class with specific parameters:
  • od_strategy (enum, ProcessorConfig, default: SNAP_ALL_SIDE): Strategy to align geodata in open domain (not covered by reference polygons).
  • threshold_overlap_percentage (%)(0-100) (default 50)
  • crs: The Coordinate Reference System (CRS) (default: EPSG:31370 - Belgian Lambert72)
• Load thematic data
• Load reference data
• Process (align) the thematic data with relevant_distance values passed to process(...), predict(...) or evaluate(...)
• Results are returned:
  • Resulting geometry
  • Differences: parts that are 'different' from the original geometry (positive or negative)
  • Positive differences: parts that are added to the original geometry
  • Negative differences: parts that are removed form the original geometry
  • Relevant intersections: relevant intersecting parts of the reference geometries
  • Relevant differences: relevant differences of the reference geometries

The brdr-algorithm

The algorithm for alignment is based on 2 main principles:

• Principle of intentionality: Thematic boundaries can consciously or unconsciously deviate from the reference borders. The algorithm should keep notice of that.
• Selective spatial conservation of shape: The resulting geometry should re-use the shape of the reference borders where aligned is of relevance.

The figure below shows a schematic overview of the algorithm:

The algorithm can be split into 3 main phases:

• Initialisation:
  • Deciding which reference polygons are candidate-polygons to re-use its shape. The reference candidate polygons are selected based on spatial intersection with the thematic geometry.
• Processing:
  • Process all candidate-reference polygons one-by-one
  • Calculate relevant zones for each candidate-reference-polygon
    • relevant intersections: zones that must be present in the final result
    • relevant differences: zones that must be excluded from the final result
  • Evaluate each candidate based on their relative zones: which parts must be kept and which parts must be excluded
  • Union all kept parts to recompose a resulting geometry
• Post-processing:
  • Validation/correction of differences between the original input geometry and the composed intermediate resulting geometry after processing the algorithm
  • Technical validation of inner holes and multipolygons that are created by processing the algorithm
  • Clean-up slivers
  • Make the resulting geometry valid

RESULT:

A new resulting output geometry, aligned to the reference-polygons

Development

pip-compile

PIP_COMPILE_ARGS="-v --strip-extras --no-header --resolver=backtracking --no-emit-options --no-emit-find-links"
pip-compile $PIP_COMPILE_ARGS
pip-compile $PIP_COMPILE_ARGS -o requirements-dev.txt --all-extras

tests

python - m
pytest - -cov = brdr
tests / --cov - report
term - missing

Docker

As an example-usage (proof-of-concept), a Dockerfile is created to set up a demo - mapviewer with a BRDR webservice that 'predicts' one or multiple actual geometries for a input-geometry This webservice uses 'brdr' to do the calculations/predictions.

This POC can be found at brdr-webservice.

Motivation & citation

A more in-depth description of the algorithm can be found in the following article (in dutch):

• Dieussaert, K., Vanvinckenroye, M., Vermeyen, M., & Van Daele, K. (2024). Grenzen verleggen. Automatische correcties van geografische afbakeningen op verschuivende onderlagen Onderzoeksrapporten Agentschap Onroerend Erfgoed, 332. https://doi.org/10.55465/SXCW6218.

Comments and contributions

We would love to hear from you and your experiences with brdr or its sister project brdrQ. The discussions forum is the place to be when:

• You have any questions on using brdr or brdrQ or their applicability to your use cases
• Want to share your experiences with the library
• Have any suggestions for improvements or feature requests

If you have discovered a bug in the brdr library you can report it here:

https://github.com/OnroerendErfgoed/brdr/issues

We try to keep the list of issues as clean as possible. If you're unsure whether something is a bug, or whether the bug is in brdr or brdrQ, we encourage you to go through the discussions forum first.

Acknowledgement

This software was created by Athumi, the Flemish data utility company, and Flanders Heritage Agency.