A Hybrid SVD–Log Ratio Framework for Satellite-Based Disaster Damage Segmentation and Impact Mapping
Indonesian Journal of Data Risk Research · 15.06.2026 · 1 (1) · 26-43
DOI: 10.55749/ijdrr.v1i1.186 ↗Kurz gefasst
Fachartikel in der Zeitschrift Indonesian Journal of Data Risk Research, online erschienen am 15.06.2026. Verfasst von Faris Alaudin Shalih und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Indonesia is a region prone to various natural disasters, requiring a monitoring system capable of detecting changes in impacted areas quickly and accurately. This study proposes a method for segmenting satellite images that combines singular value decomposition and log ratio approaches to identify disaster impacts using Sentinel-2 satellite images. The segmentation process consists of preprocessing steps such as RGB to HSV colour conversion and histogram matching. The processing stage involves applying singular value decomposition on each 2×2 kernel block, where the smallest singular value is used as an indicator of local differences. This is then combined with a log operation to reduce speckle noise. The segmentation results are refined through postprocessing by overlaying satellite data to enhance the visibility of affected areas. Quantitative evaluation using ROC curves and AUC shows that this combination achieves high detection accuracy, with a maximum AUC value of 0.9216, outperforming the Otsu method (0.8123), Region Growing (0.8349), and K-Means (0.8731). This combination has proven effective in distinguishing image changes in detail and holds potential for application in an automatic, real-time disaster impact monitoring system in disaster-prone areas.
Zitieren
APA
Shalih, F. A., & Abdurrazzaq, A. (2026). A Hybrid SVD–Log Ratio Framework for Satellite-Based Disaster Damage Segmentation and Impact Mapping. Indonesian Journal of Data Risk Research, 1(1), 26-43. https://doi.org/10.55749/ijdrr.v1i1.186
Vancouver
Shalih FA, Abdurrazzaq A. A Hybrid SVD–Log Ratio Framework for Satellite-Based Disaster Damage Segmentation and Impact Mapping. Indonesian Journal of Data Risk Research. 2026;1(1):26-43. doi:10.55749/ijdrr.v1i1.186
BibTeX
@article{Shalih2026,
author = {Shalih, Faris Alaudin and Abdurrazzaq, Achmad},
title = {A Hybrid SVD–Log Ratio Framework for Satellite-Based Disaster Damage Segmentation and Impact Mapping},
journal = {Indonesian Journal of Data Risk Research},
year = {2026},
volume = {1},
number = {1},
pages = {26--43},
doi = {10.55749/ijdrr.v1i1.186}
}
EUVIMED