startinpy

A library for modelling and processing 2.5D terrains using a (2D) Delaunay triangulation. The triangulation is computed in 2D, but the z-elevation of the vertices are kept.

The underlying code is written in Rust (so it's rather fast) and robust arithmetic is used (so it shouldn't crash). startinpy uses the startin Rust library and adds several utilities and functions, for instance NumPy support for input/output, exporting to several formats, and easy-of-use.

startinpy allows you to:

1. insert incrementally points
2. delete vertices (useful for simplification, interpolation, and other operations)
3. interpolate with several methods: TIN, natural neighbours, IDW, Laplace, etc.
4. use other useful terrain Python libraries that are also NumPy-based, eg laspy, rasterio, meshio
5. output the TIN to several formats: OBJ, PLY, GeoJSON, and CityJSON
6. store extra attributes with the vertices (the ones from LAS/LAZ)

Documentation

https://startinpy.rtfd.io

Installation

pip

To install the latest release: pip install startinpy

(watch out: this does not work with Linux currently, it installs an old version!)

If you want to compile it yourself

1. install latest Rust
2. install maturin
3. maturin build --release
4. cd ./target/wheels/
5. pip install [name-wheel].whl will install it to your local Python

Development

1. install Rust (v1.39+)
2. install maturin
3. maturin develop
4. move to another folder, and import startinpy shouldn't return any error

Testing

To run the automated test suite:

1. install the test requirements: pip install -r tests/requirements.txt
2. pytest

Examples

The folder ./demo contains a few examples.

import laspy
import numpy as np
import startinpy

las = laspy.read("../data/small.laz")
pts = np.vstack((las.x, las.y, las.z)).transpose()

dt = startinpy.DT()
dt.insert(pts)

# -- remove vertex #4
try:
    dt.remove(4)
except Exception as e:
    print(e)

print("# vertices:", dt.number_of_vertices())
print("# triangles:", dt.number_of_triangles())

# -- print the vertices forming the convex hull, in CCW-order
print("CH: ", dt.convex_hull())

# -- fetch all the incident triangles (CCW-ordered) to the vertex #235
vi = 235
one_random_pt = dt.points[vi]
print("one random point:", one_random_pt)
print(dt.incident_triangles_to_vertex(vi))

# -- interpolate at a location with the linear in TIN method
zhat = dt.interpolate({"method": "TIN"}, [[85718.5, 447211.6]])
print("result: ", zhat[0])

If you use this software, please cite this article

@article{Ledoux24,
  author = {Ledoux, Hugo},
  title = {{startinpy}: {A} {P}ython library for modelling and processing {2.5D} triangulated terrains},
  journal = {Journal of Open Source Software},
  year = {2024},
  volume = {9},
  number = {103},
  pages = {7123},
  doi = {10.21105/joss.07123}
}