AROSICS is an automated and robust open-source image co-registration software for multi-sensor satellite data.
AROSICS is an automated and robust open-source image co-registration software for multi-sensor satellite data. The Python package performs automatic subpixel co-registration of two satellite image datasets based on an image matching approach working in the frequency domain, combined with a multistage workflow for effective detection of false-positives.
AROSICS was developed at the German Research Centre for Geosciences Potsdam (GFZ) in the context of GeoMultiSens. The underlying algorithm has been published in Scheffler et al. 2017:
The package is freely available from the Python package index or conda-forge and is already widely used by a community of geospatial analysts and remote sensing experts.
The source code can be accessed at the GitLab repository or its mirror repository on GitHub. Documentation is available here.
AROSICS detects and corrects local as well as global misregistrations between two input images in the subpixel scale, that are often present in satellite imagery. The algorithm is robust against the typical difficulties of multi-sensoral/multi-temporal images. Clouds are automatically handled by the implemented outlier detection algorithms. The user may provide masks to exclude certain image areas from tie point creation. The image overlap area is automatically detected. AROSICS supports a wide range of input data formats and can be used from the command line (without any Python experience) or as a normal Python package.
Two co-registration modes are available: the local co-registration approach accounts for locally varying shifts and computes hundreds of tie points spread over the entire image overlap whereas the global co-registration approach only corrects a static, translational X/Y shift.
AROSICS provides a lot of functionality to visualize the initial mis-registration and to create interactive plots in a Jupyter notebook. The figure shown here, e.g., visualizes the computed shift vectors (local co-registration approach) after filtering false-positives, mainly due to clouds in the target image.
The development of AROSICS was funded by the German Federal Ministry of Education and Research (project grant code: 01 IS 14 010 A-C).
All images are co-registered using the high resolution reference data. Among the several algorithms, a python package working in the frequency domain, namely Automated and Robust Open-Source Image Co-Registration Software (AROSICS), is adopted owing to its practical capabilities, such as ensuring tie point data export, including displacement vector information, and easy implementation of batch processing thanks to the Python environment (Scheffler et al. 2017).
A subsequent coregistration is therefore required. For this, the automatic subpixel coregistration package arosics by [48] is used. Its image matching approach within the frequency domain, outlier detection algorithms for robustness and design around remote sensing data have shown to yield the best results.
Subsequently, we used AROSICS (Automated and Robust Open-Source Image Co-Registration Software) to automatically detect and correct local geometric biases between multi-temporal GF-6 images, so that the spatial error was controlled within 1 pixel (Scheffler et al., 2017).
In the case of AROSICS, the estimated offsets are better than 1/10th of a pixel. As mentioned earlier, some parameters can be tweaked to get better results, but again, that requires extensive experimentation with these methods.
AROSICS (Automated and Robust Open-Source Image Co-Registration Software) is a significant advancement in geospatial data processing. It employs an approach based on image matching in the frequency domain, along with a multistage workflow [...], to efficiently recognize and eliminate false positives. [...] AROSICS demonstrates remarkable robustness, making it well suited for large datasets and challenging distortion scenarios. [...] AROSICS is used for both Sentinel-2 and Planet-Scope images.
We co-registered the different sensor products following Scheffler et al. (2017). This allowed stacking of imagery with near pixel-perfect overlap of ant mound locations in different bands.
The Environmental Mapping and Analysis Program (EnMAP) is a German hyperspectral satellite mission that aims at monitoring and characterising Earth’s environment on a global scale. EnMAP measures and models key dynamic processes of Earth’s ecosystems by extracting geochemical, biochemical and biophy
SpecHomo is a Python package for spectral homogenization of multispectral satellite data, i.e., for the transformation of the spectral information of one sensor into the spectral domain of another one.