Skip to main content

cdBoundary

A very fast 2.5D concave hull algorithm by distance implemented in Python.

Rather than using a factor between 0 and 1 as most Concave Hull or Alpha Shape algorithms this implementation uses maximum edge length (tol) as a paramater.

The library was needed specific for survey or LiDAR data for engineering applications.

Install

git clone https://github.com/civildot/cdBoundary
cd cdBoundary
python setup.py install

Usage

1. As a library

The points should be in list format with coordinate pairs:
[[x1, y1], [x2, y2], [x3, y3], ...] or
[[x1, y1, z1], [x2, y2, z2], [x3, y3, z3], ...]

All edge lengths exceeding tol parameter will be attempted to be removed. This parameter’s unit will thus be the same as your coordinates. Wether it is meters, feet or degrees.

The ../examples/points-1k.json dataset was borrowed with thanks from concaveman.

import json
from cdBoundary.boundary import ConcaveHull

ch = ConcaveHull()
with open('../examples/points-1k.json') as source:
    pts = json.load(source)
ch.loadpoints(pts)
ch.calculatehull(tol=0.0025)

The result is a shapely Polygon at ch.hull.

print('Geometry type:', ch.hull.geom_type)
coords = ch.boundary_points()
print()
print('The first 5 coordinates of the concave hull is:')
for i in range(5):
    print(i+1, '.', coords[i], sep='')
Geometry type: Polygon

The first 5 coordinates of the concave hull is:
1.(-122.08441, 37.384634)
2.(-122.0832, 37.383161)
3.(-122.08214, 37.3812)
4.(-122.08204, 37.380184)
5.(-122.08216, 37.379173)
ch.plot()

If there is an uncertianty on which value to use for tol the estimate function can be used to provide an estimate. The default percentile perc is 92.5%. At this value only outside border edges longer than 92.5% of all the edges will be attempted to be removed.

print('There is', len(coords), 'vertices on the current polygon.')
print('The suggested value to use is', round(ch.estimate(), 10))
ch.calculatehull(tol=ch.estimate())
print()
print('There is', len(ch.boundary_points()), 'vertices on the new polygon.')
There is 83 vertices on the current polygon.
The suggested value to use is 0.0017366632

There is 164 vertices on the new polygon.

The low value for the estimate() (much smaller than 1) suggest that the coordinates are in degress. When working in feet or meters it will be much larger.

ch.plot()

2. Working directly with files

Fiona is used to read point data directly from files. The resulting Polygon can also be written to the file format supported by Fiona.

from cdBoundary.fileio import FileIO

fch = FileIO()
fch.file2points('../examples/Bandelierkop_survey.shp')
fch.write2file(tol=35)

In the same folder as your input file there is now a file named concave_hull.shp (default) containing the concave hull as a polygon.

or

with open('../examples/points-1k.json') as source:
    pts = json.load(source)
fch = FileIO()
fch.loadpoints(pts)
fch.calculatehull(tol=ch.estimate())
fch.write2file(outfile='../examples/concave_hull2.shp', driver='ESRI Shapefile')

Documentation

https://civildot.github.io/cdBoundary

Requirements / Dependencies

Algorithm

This concave hull implementation is based on the algorithm developed by Duckham et al. (2008) in the paper “Efficient generation of simple polygons for characterizing the shape of a set of points in the plane”, available here.

Wish list

Possible features or enhancements in the future. Nothing planned. The library is already satisfactory for our speicific application.

  • Plot output for quick visualization Implemented 15 July 2023
  • Support an elevation (Z value) on the polygon vertices Implemented 15 July 2023
  • Support for holes
  • cli (Command Line Interface)
  • Provide as a QGIS Plugin
  • Implement it as a Streamlit application

Similar Implementations

A lot of concave-hull and alpha-shape implementations on Github, but very few using maximum edge length as an argument.

  1. The greatest influence on cdBoundary is http://www.rotefabrik.free.fr/concave_hull which can be accessed through the OpenJUMP user interface.
  2. The lasboundary command line program from LASTools is unfortunately proprierity but also edge length as an argument and is very powerful.
  3. MacRoad, later HighRoad, an ancient road design and earthworks program. Still being used today but no longer actively maintained. It has a Join Edge Points function which is a concave hull implementation already implemented in 1988. It has an estimate button which inspired the estimate function for cdBoundary.
  4. jtsop from the JTS (Java Topology Suite) has got a Construction.concaveHullByLen function as well.

Metadata

Release files for cdBoundary 0.8

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for cdBoundary 0.8
File Size Uploaded
cdBoundary-0.8.tar.gz 12.2 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for cdBoundary 0.8
File Interpreter ABI Platform
cdBoundary-0.8-py3-none-any.whl Python 3 none any Details

Total release size: 21.7 kB

Release files / cdBoundary-0.8.tar.gz

Download URL cdBoundary-0.8.tar.gz
Size 12.2 kB
Tags Source
SHA-256 checksum
How to use checksums
b642181fa5934befe80d6a2c86a443ca5cb13ce3cf7c4d903beb6680f7135e7e
BLAKE2b-256 checksum
How to use checksums
d3f0b33d53692edf2dd89205189c5e04290c4d6d1b7ae4f131b70f6ac1efb748
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/4.0.2 CPython/3.11.3

Release files / cdBoundary-0.8-py3-none-any.whl

Download URL cdBoundary-0.8-py3-none-any.whl
Size 9.5 kB
Tags Python 3
SHA-256 checksum
How to use checksums
6142658a3bad7f8d597aca3d9c0d93c28bb162079ea5e3ed9027fd3497d38538
BLAKE2b-256 checksum
How to use checksums
26eb8b18022be46bce216efcfaab63de09adeb1099a464703c7ebe68f7f05bc2
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/4.0.2 CPython/3.11.3

Release history Release notifications | RSS feed

This release

0.8 This release

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page