Abstract

We propose a method for content-based retrieval of solar magnetograms using semantic hashing. The approach is based on data from the Helioseismic and Magnetic Imager (HMI) aboard the Solar Dynamics Observatory (SDO), processed using the SunPy and PyTorch libraries. A mathematical representation of the Sun’s magnetic field regions is constructed in the form of a fixed-length vector, enabling efficient similarity comparisons without the need to analyze full-disk images directly. To reduce retrieval time and dimensionality, a fully connected autoencoder is employed to compress a 400-dimensional descriptor into a compact 50-dimensional semantic hash. Experimental results demonstrate the effectiveness of the proposed approach, achieving the highest precision among evaluated state-of-the-art methods. In addition to image retrieval, the method is also applicable to solar image classification tasks.

Recommended Citation

Grycuk, R. & Scherer, R. (2025). Enhancing Solar Magnetogram Retrieval with Deep Semantic Hashing and Hierarchical Graph IndexingIn I. Luković, S. Bjeladinović, B. Delibašić, D. Barać, N. Iivari, E. Insfran, M. Lang, H. Linger, & C. Schneider (Eds.), Empowering the Interdisciplinary Role of ISD in Addressing Contemporary Issues in Digital Transformation: How Data Science and Generative AI Contributes to ISD (ISD2025 Proceedings). Belgrade, Serbia: University of Gdańsk, Department of Business Informatics & University of Belgrade, Faculty of Organizational Sciences. ISBN: 978-83-972632-1-5. https://doi.org/10.62036/ISD.2025.45

Paper Type

Poster

DOI

10.62036/ISD.2025.45

Share

COinS
 

Enhancing Solar Magnetogram Retrieval with Deep Semantic Hashing and Hierarchical Graph Indexing

We propose a method for content-based retrieval of solar magnetograms using semantic hashing. The approach is based on data from the Helioseismic and Magnetic Imager (HMI) aboard the Solar Dynamics Observatory (SDO), processed using the SunPy and PyTorch libraries. A mathematical representation of the Sun’s magnetic field regions is constructed in the form of a fixed-length vector, enabling efficient similarity comparisons without the need to analyze full-disk images directly. To reduce retrieval time and dimensionality, a fully connected autoencoder is employed to compress a 400-dimensional descriptor into a compact 50-dimensional semantic hash. Experimental results demonstrate the effectiveness of the proposed approach, achieving the highest precision among evaluated state-of-the-art methods. In addition to image retrieval, the method is also applicable to solar image classification tasks.