Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods

Ufuk Kaya; Behcet Ugur Toreyin; Fatma Kucuk

doi:10.1109/MIGARS67156.2025.11232406

Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods

Ufuk Kaya^*
, Behcet Ugur Toreyin
, Fatma Kucuk

^*Corresponding author for this work

Informatics Institute

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

This paper proposes a hybrid anomaly detection framework for hyperspectral images by combining Robust Sub-space Recovery (RoSuRe) and Local Anomaly Detection (LAD). Both RoSuRe and its variant Fixed Dictionary RoSuRe (FD-RoSuRe) are evaluated with LAD-C and LAD-C-S across multiple datasets. Experiments with heatmaps, ROC curves, and AUC scores show that FD-RoSuRe with LAD-C-S yields the most accurate and localized detection, especially in structured and cluttered environments.

Original language	English
Title of host publication	2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9798331579203
DOIs	https://doi.org/10.1109/MIGARS67156.2025.11232406
Publication status	Published - 2025
Event	3rd International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025 - Bucharest, Romania Duration: 2 Sept 2025 → 4 Sept 2025

Publication series

Name	2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025

Conference

Conference	3rd International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025
Country/Territory	Romania
City	Bucharest
Period	2/09/25 → 4/09/25

Bibliographical note

Publisher Copyright:
© 2025 IEEE.

Access to Document

10.1109/MIGARS67156.2025.11232406

Cite this

Kaya, U., Toreyin, B. U., & Kucuk, F. (2025). Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods. In 2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025 (2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/MIGARS67156.2025.11232406

Kaya, Ufuk ; Toreyin, Behcet Ugur ; Kucuk, Fatma. / Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods. 2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025. Institute of Electrical and Electronics Engineers Inc., 2025. (2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025).

@inproceedings{36874ee0a27246da9fabe8d38121a33a,

title = "Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods",

abstract = "This paper proposes a hybrid anomaly detection framework for hyperspectral images by combining Robust Sub-space Recovery (RoSuRe) and Local Anomaly Detection (LAD). Both RoSuRe and its variant Fixed Dictionary RoSuRe (FD-RoSuRe) are evaluated with LAD-C and LAD-C-S across multiple datasets. Experiments with heatmaps, ROC curves, and AUC scores show that FD-RoSuRe with LAD-C-S yields the most accurate and localized detection, especially in structured and cluttered environments.",

author = "Ufuk Kaya and Toreyin, \{Behcet Ugur\} and Fatma Kucuk",

note = "Publisher Copyright: {\textcopyright} 2025 IEEE.; 3rd International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025 ; Conference date: 02-09-2025 Through 04-09-2025",

year = "2025",

doi = "10.1109/MIGARS67156.2025.11232406",

language = "English",

series = "2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025",

address = "United States",

}

Kaya, U, Toreyin, BU & Kucuk, F 2025, Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods. in 2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025. 2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025, Institute of Electrical and Electronics Engineers Inc., 3rd International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025, Bucharest, Romania, 2/09/25. https://doi.org/10.1109/MIGARS67156.2025.11232406

Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods. / Kaya, Ufuk; Toreyin, Behcet Ugur; Kucuk, Fatma.
2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025. Institute of Electrical and Electronics Engineers Inc., 2025. (2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods

AU - Kaya, Ufuk

AU - Toreyin, Behcet Ugur

AU - Kucuk, Fatma

PY - 2025

Y1 - 2025

N2 - This paper proposes a hybrid anomaly detection framework for hyperspectral images by combining Robust Sub-space Recovery (RoSuRe) and Local Anomaly Detection (LAD). Both RoSuRe and its variant Fixed Dictionary RoSuRe (FD-RoSuRe) are evaluated with LAD-C and LAD-C-S across multiple datasets. Experiments with heatmaps, ROC curves, and AUC scores show that FD-RoSuRe with LAD-C-S yields the most accurate and localized detection, especially in structured and cluttered environments.

AB - This paper proposes a hybrid anomaly detection framework for hyperspectral images by combining Robust Sub-space Recovery (RoSuRe) and Local Anomaly Detection (LAD). Both RoSuRe and its variant Fixed Dictionary RoSuRe (FD-RoSuRe) are evaluated with LAD-C and LAD-C-S across multiple datasets. Experiments with heatmaps, ROC curves, and AUC scores show that FD-RoSuRe with LAD-C-S yields the most accurate and localized detection, especially in structured and cluttered environments.

UR - https://www.scopus.com/pages/publications/105025423505

U2 - 10.1109/MIGARS67156.2025.11232406

DO - 10.1109/MIGARS67156.2025.11232406

M3 - Conference contribution

AN - SCOPUS:105025423505

T3 - 2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025

BT - 2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 3rd International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025

Y2 - 2 September 2025 through 4 September 2025

ER -

Kaya U, Toreyin BU, Kucuk F. Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods. In 2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025. Institute of Electrical and Electronics Engineers Inc. 2025. (2025 International Conference on Machine Intelligence for GeoAnalytics and Remote Sensing, MIGARS 2025). doi: 10.1109/MIGARS67156.2025.11232406

Hybrid Hyperspectral Anomaly Detection via Robust Subspace Recovery and Laplacian Cauchy-Based Methods

Abstract

Publication series

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this