Filter geolocalized photo sequences. Useful to move unuseful images before uploading sequences on Panoramax.
usage: photo_filter.py [-h] [-d DUPLICATE_DISTANCE] [-j JSON_FILE] [-r]
[-t MAX_TURN_ANGLE] [-e ENCLOSING_IMAGES] [-v]
paths [paths ...]
- Search for distance based duplicate images and move them in a 'duplicate' subfolder.
- Search for image in geofence zones and move them in a 'geofence' subfolder
- Search for images with a too large angle between them.
positional arguments:
paths paths to the images folders
optional arguments:
-h, --help show this help message and exit
-r, --recursive search images in subdirectory
-d, --duplicate_distance DUPLICATE_DISTANCE
Min distance in meter for duplicate image detection.
Default: 0.5 meter
-j, --json_file JSON_FILE
Path to the geojson file containing geofence polygons
-t, --max_turn_angle MAX_TURN_ANGLE
Check if two subsequent images have a too large direction angle.
The result will be send to Josm as an image layer.
Default: False
-e, --enclosing_images ENCLOSING_IMAGES
Set how many images will be added around a too tight angle. These images will be included in the Josm session.
Exemple with 10: 2 images before and 8 images after.
Default: 10 images
-v, --version show program's version number and exit
- Create a virtual environnement:
python3 -m venv photo_filter - Move to this environnement:
cd photo_filter - Activate this environnement: linux:
source bin/activate- windows powershell:bin\script\activate.ps1 - Clone this repository:
git clone https://github.com/Stefal/Photo_filter.git - Install the requirements:
pip install -r Photo_filter/requirements.txt
I usually create the geojson file with Josm.
- Add an empty layer
- Create a closed polygon around the area you want to exclude
- tag them with an area key/value. I use
landuse=military - export as geojson