Python PointSet Class

This package includes a class for handling point-coordinates and their datum using EPSG codes. For datum transformations, this package makes use of the PyProj package. It meant to simplify coordinate transformations between EPSG codes.

Installation

python3 -m pip install pointset

Usage

The PointSet class wraps pyproj in order to allow coordinate transformations. In the following, the functionality of the class is explained.

PointSet with Datum information

In the example above we just generated random positions without any datum information. However the main feature of this class is datum transformations.

First, we define some points in UTM 32N (EPSG: 25832)

xyz_utm = np.array(
    [
        [364938.4000, 5621690.5000, 110.0000],
        [364895.2146, 5621150.5605, 107.4668],
        [364834.6853, 5621114.0750, 108.1602],
        [364783.4349, 5621127.6695, 108.2684],
        [364793.5793, 5621220.9659, 108.1232],
        [364868.9891, 5621310.2283, 107.9929],
        [364937.1665, 5621232.2154, 107.9581],
        [364919.0140, 5621153.6880, 107.8130],
        [364906.8750, 5621199.2600, 108.0610],
        [364951.9350, 5621243.4890, 106.9560],
        [364992.5600, 5621229.7440, 106.7330],
        [365003.7740, 5621203.8200, 106.7760],
        [364987.8850, 5621179.5160, 107.8890],
        [364950.1180, 5621148.5770, 107.9120],
    ]
)

utm_point_set = PointSet(xyz=xyz_utm, epsg=25832)

transform point set to another EPSG:

utm_point_set.to_epsg(4936)
print(utm_point_set.mean())

Output:

EPSG: 4936
Coordinates:
[[4014743.91215813  499064.14065106 4914468.86763503]]

As you can see, the coordinates are now given in a global cartesian frame (EPSG: 4936).

Local Coordinate Frame

You can transform the coordinates of a pointset in a local ellipsoidal coordinate frame tangential to the GRS80 ellipsoid for local investigations using .to_local() or .to_epsg(0)

utm_point_set.to_local()
print(utm_point_set.mean())

Output:

EPSG: 0
Coordinates:
[[-6.15055943e-11 -5.61509442e-10  2.01357074e-10]]

Note, that the mean of the PointSet will be zero in local coordinates. Internally, a local_transformer object is created, that takes care of the transformation to local coordinates. Especially for comparing PointSets, it might be useful to analyze both PointSets in the same local coordinate frame. You can do this by setting the local_transformer variable either during instance creation or later:

point_set.local_transformer = utm_point_set.local_transformer
point_set = PointSet(xyz=xyz, epsg=0, local_transformer=utm_point_set.local_transformer)

Now, the newly created pointset has the same datum information as the utm-coordinates.

PointSet without Datum information

Define PointSet with random numbers

from pointset import PointSet

xyz = np.random.randn(10000, 3) * 20
point_set = PointSet(xyz=xyz)

print the point set to see the EPSG code and the coordinates:

print(point_set)

Output (for example):

EPSG: 0
Coordinates:
[[  2.61185114  26.86022378  24.16762049]
 [-13.10880044  -0.59031669  25.03318095]
 [ 11.7225511   -8.60815889   8.14436657]
 ...
 [  2.92442258 -24.89119898  -2.17729086]
 [  1.45229968  24.66663312  21.73038683]
 [ 15.90327212  28.88909949   4.56549931]]

Because we only provided the numpy array and no other parameters, this PointSet has no datum information. The positions are assumed to be in a local unknown frame, which is denoted with an EPSG code of 0. Therefore, we will get an error if we try to change the EPSG code to some global frame, e.g. EPSG: 4937

try:
    point_set.to_epsg(4937)
except PointSetError as e:
    print(e)

Output:

Unable to recover from local frame since definition is unknown!

However, we can still do some data operations like computing the mean of the point_set (which will also be a PointSet):

mean_pos = point_set.mean()

We can access the raw values of the PointSet coordinates using x y z or xyz

print(f"Mean: {mean_pos.xyz}, x = {mean_pos.x:.3f}, y = {mean_pos.y:.3f}, z = {mean_pos.z:.3f}")

Output:

Mean: [[-0.03867169  0.21157332  0.0836462 ]], x = -0.039, y = 0.212, z = 0.084

It also possible to change the values in this way:

mean_pos.y = 10
print(f"Changed y-value to: {mean_pos.y:.3f}")

Output:

Changed y-value to: 10.000

To add / substract two PointSets, use normal operators:

added_point_set = point_set + mean_pos

You can create a deep copy of the PointSet using .copy():

copied_point_set = point_set.copy()