AUTOGRAF (AUTomated Orthorectification of GRAFfiti photos) is an open-source Python-based Metashape add-on that enables the automated orthorectification of graffiti photos at a specific site of interest. It employs state-of-the-art photogrammetric computer vision techniques to allow highly accurate georeferencing and orthorectification of large numbers of photographs. A paper published in the MDP journal Heritage details AUTOGRAF's methodology.
AUTOGRAF is developed as part of the INDIGO project (In-ventory and DI-sseminate G-raffiti along the d-O-naukanal) carried out by the Ludwig Boltzmann Institute for Archaeological Prospection and Virtual Archaeology in close collaboration with the GEO Department of TU Wien University.
Before AUTOGRAF can be used, the following preparatory steps [1-3] need to be performed:
Agisoft's Metashape version 1.8.3 (earlier versions might not work as AUTOGRAF was designed and tested on 1.8.3) must be installed, and the license must be active. More info here: https://www.agisoft.com/downloads/installer/
Some external Python packages must installed into Metashape's Python environment. These packages are: numpy, matplotlib, scikit-image. To do this the following command must be executed via the command line (note that programme paths might need to be adapted)
%programfiles%\Agisoft\Metashape Pro\python\python.exe" -m pip install numpy matplotlib scikit-image
/MetashapePro.app/Contents/Frameworks/Python.framework/Versions/3.8/bin/python3.8 -m pip install numpy matplotlib scikit-image
./metashape-pro/python/bin/python3.8 -m pip install numpy matplotlib scikit-image
First, you have to download the AUTOGRAF.py script from this repository (./src/AUTOGRAF.py). Then you must add it to the METASHAPE GUI. There are several ways to do this. For example, as follows (a or b):
a) In the Metahsape GUI open Tools -> Run Script -> select the AUTOGRAF.py
b) If you use AUTOGRAF regularly, you can also add it permanently to Metashape by copying AUTOGRAF.py to the following directory (this is a Windows example, MacOS and Linux are similar):
C:/Users/<username>/AppData/Local/Agisoft/Metashape Pro/scripts/
Both (a and b) should result in a GUI looking like this:
AUTOGRAF follows the methodology introduced by Wild et al. (in preparation)1. The general workflow is summarised in the following workflow chart1:
A prerequisite of the currently implemented version of AUTOGRAF is that a network of oriented cameras exists. This network must be acquired at the studied site prior to using AUTOGRAF as AUTOGRAF uses the existing network to incrementally orient and add new photographs. In Metashape terminology this network is realised through a "main" chunk containing all oriented images. IMPORTANT: this "main" chunk must be the first chunk in the list of chunks:
As input, AUTOGRAF expects one folder, containing all subfolders that need to be processed. A subfolder contains the images of one graffito:
AUTOGRAF is started by clicking the "Run" button in the dropdown menu. AUTOGRAF automatically executes the above methodology and writes the resulting orthophotos in a "results" folder of each subfolder. AUTOGRAF produces 2 main results:
- Orthophotos in different raster cell sizes (1cm / 1mm / native)
- A custom-made georeferencing file (.csv). This file contains the transformation parameters that allow assigning a 3D world coordinate to each pixel. Please note that this file is not standardised (as no standard for this exists yet)
Currently, AUTOGRAF still has the following limitations:
- only Metashape version 1.8 is supported;
- it only supports rectilinear lenses (e.g. fisheye lenses are NOT supported yet);
- only one coordinate reference system is supported (MGI/Austria GK East, EPSG:31256; https://epsg.io/31256).
All those things can be easily altered in the code for your own use, but we might consider implementing the above-mentioned features in the future (your feedback is thus highly appreciated!) Please help us improve AUTOGRAF by raising an issue here on GitHub or directly writing an E-Mail to bewild@projectindigo.eu
For citing the methodology, please use this reference:
1Wild, B., Verhoeven, G., Schlegel, J., Wogrin, S., Wieser, M., Ressl, C., Otepka-Schremmer, J., Pfeifer, N. 2022. AUTOGRAF - AUTomated Orthorectification of GRAFfiti photos. Heritage, 5(4), 2987–3009. https://doi.org/10.3390/heritage5040155
If you use the provided code, please also cite AUTOGRAF as follows:
Wild B. 2022. AUTOGRAF (AUTomated Orthorectification of GRAFfiti photos). Zenodo. https://doi.org/10.5281/zenodo.7049950