https://github.com/jmigueldelgado/fake-geo-images

A module to programmatically create geotiff images which can be used for unit tests
https://github.com/jmigueldelgado/fake-geo-images

geospatial image-processing p2 processing testing unit-testing

Last synced: 16 days ago
JSON representation

A module to programmatically create geotiff images which can be used for unit tests

• Host: GitHub
• URL: https://github.com/jmigueldelgado/fake-geo-images
• Owner: jmigueldelgado
• License: mit
• Created: 2020-10-01T15:04:15.000Z (about 4 years ago)
• Default Branch: master
• Last Pushed: 2024-08-20T06:50:43.000Z (3 months ago)
• Last Synced: 2024-09-19T08:47:58.567Z (about 2 months ago)
• Topics: geospatial, image-processing, p2, processing, testing, unit-testing
• Language: Python
• Homepage:
• Size: 43 KB
• Stars: 27
• Watchers: 11
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# 🚧 fake-geo-images

[WARNING] The repo's ownership is currently being transfered. It might take a few days until the new owner has finalized the setup.

A module to programmatically create geotiff images which can be used for unit tests.



    



The underlying idea is that in order to write unit tests for geospatial image processsing algorithms, 

it is necessary to have an actual input image file or array. Organising these test images becomes a chore over time,

they should not be stored in git as they are large binary data and when stored outside, there always

is the danger that they are not updated according to changes in the code repo.

**fake-geo-images** provides a solution to the problem by providing simple code that allows to create

geospatial images (so far geotiffs) in a parameterised way. 

## Install package

```bash

pip install fake-geo-images

```

## Run tests

```bash

pytest

```

## Usage

In the following an example unit test for a hypothetical NDVI function.

```python

import numpy as np

import rasterio as rio

from pathlib import Path

from rasterio.transform import from_origin

from my_image_processing import ndvi

from fake_geo_images.fakegeoimages import FakeGeoImage

def test_ndvi():

    """

    A unit test if an NDVI method works in general

    """

    # Create 4-band image simulating RGBN as needed for NDVI

    test_image, _ = FakeGeoImage(

        300,

        150,

        4,

        "uint16",

        out_dir=Path("/tmp"),

        crs=4326,

        nodata=0,

        nodata_fill=3,

        cog=False,

    ).create(seed=42, transform=from_origin(13.428596, 52.494384, 0.000006, 0.000006))

    ndvi_image = ndvi(test_image)

    with rio.open(str(ndvi_image)) as src:

        ndvi_array = src.read()

        # NDVI only has one band of same size as input bands

        assert ndvi_array.shape == (1, 300, 150)

        # NDVI has float values between -1 and 1

        assert ndvi_array.dtype == np.float

        assert ndvi_array.min >= -1

        assert ndvi_array.max <= 1

```