DSUPPOSe_Tiles: an accelerated tile-based super-resolution method for full-field image processing with an open-source implementation
Journal of the Optical Society of America B · 01.05.2026 · 43 (5) · 888
DOI: 10.1364/josab.592662 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of the Optical Society of America B, online erschienen am 13.04.2026. Verfasst von Micaela Toscani und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 37 Literaturangaben. Gefördert von: Agencia Nacional de Promoción de la Investigación, el Desarrollo Tecnológico y la Innovación, Consejo Nacional de Investigaciones Científicas y Técnicas, Universidad de Buenos Aires.
Abstract (Originalsprache)
In this work, we introduce what is to our knowledge a new super-resolution method, DSUPPOSe_Tiles, that removes the boundary artifacts present in DSUPPOSe, allowing for the application to full-field images by using tiles. SUPPOSe and the faster variant, DSUPPOSe, address single-shot super-resolution by assuming that the unknown object can be represented as a superposition of virtual point sources of equal intensities. Since DSUPPOSe assumes that the object is localized far from the boundary of the image, if this is not fulfilled, the solution presents artifacts near the boundary due to ringing effects. DSUPPOSe_Tiles overcomes this problem by introducing a new fitting function that eliminates these artifacts, allowing full-image processing through tiling. The method was tested using CPU-based parallel processing and compared with the previous DSUPPOSe on a family of synthetic images, achieving the same accuracy (10 times better than the original resolution) while running ten times faster. Experimental images of Argolight-SIM calibration samples, as well as images obtained by photographing synthetic microtubule structures from the dataset of Laydi et al ., were tested. An initial resolution improvement factor of 3 was obtained, asymptotically approaching a value of 5 at longer computation times. In these experiments, the method reached an accuracy value approximately eight times lower than the original resolution, and it was shown that the computational time depends not only on the desired resolution improvement factor but also on the uniformity of the image intensities. As part of this work, we developed an open-source resource, which is available together with its documentation through our GitLab repository.
Förderung
- Agencia Nacional de Promoción de la Investigación, el Desarrollo Tecnológico y la Innovación
- Consejo Nacional de Investigaciones Científicas y Técnicas
- Universidad de Buenos Aires
Zitieren
APA
Toscani, M., Gómez, E., & Martínez, S. (2026). DSUPPOSe_Tiles: an accelerated tile-based super-resolution method for full-field image processing with an open-source implementation. Journal of the Optical Society of America B, 43(5), 888. https://doi.org/10.1364/josab.592662
Vancouver
Toscani M, Gómez E, Martínez S. DSUPPOSe_Tiles: an accelerated tile-based super-resolution method for full-field image processing with an open-source implementation. Journal of the Optical Society of America B. 2026;43(5):888. doi:10.1364/josab.592662
BibTeX
@article{Toscani2026,
author = {Toscani, Micaela and Gómez, Emanuel and Martínez, Sandra},
title = {DSUPPOSe_Tiles: an accelerated tile-based super-resolution method for full-field image processing with an open-source implementation},
journal = {Journal of the Optical Society of America B},
year = {2026},
volume = {43},
number = {5},
pages = {888},
doi = {10.1364/josab.592662}
}
EUVIMED