Detection of Illegal Electrical Transformers by Using Hyperspectral Image Analysis


Kütük M., Özdemir O. B., Ünveren V., Koz A., Alatan A. A.

IGARSS 2026 - 2026 IEEE International Geoscience and Remote Sensing Symposium, District Of Columbia, Amerika Birleşik Devletleri, 9 - 14 Ağustos 2026, ss.5705-5709, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/igarss60577.2026.11730263
  • Basıldığı Şehir: District Of Columbia
  • Basıldığı Ülke: Amerika Birleşik Devletleri
  • Sayfa Sayıları: ss.5705-5709
  • Orta Doğu Teknik Üniversitesi Adresli: Evet

Özet

This paper presents a UAV-based hyperspectral sensing framework for detecting illegal electrical transformers in agricultural irrigation areas in Türkiye, where on-site inspections are challenging due to distance, access limitations, and security risks. The workflow combines ground-based spectral measurements with airborne hyperspectral imaging. Representative transformer materials are first measured with an ASD spectrometer to build a spectral library capturing material-specific reflectance signatures. A UAV-mounted hyperspectral camera then acquires VNIR data (400–1000 nm, 273 bands) at 10 cm ground sampling distance from 80 m altitude. Transformer candidates are localized using Spectral Angle Mapper and Spectral Matched Filter algorithms. Field experiments in rural districts of Şanlıurfa and Kayseri analyzed twelve hyperspectral scenes containing 19 transformers. All transformers were successfully detected, with Spectral Angle Mapper achieving full detection with fewer false alarms. Both methods effectively suppressed common rural backgrounds. The study demonstrates that hyperspectral signatures provide strong discrimination between transformers and visually similar infrastructure, offering a remote, safe, and cost-efficient alternative to on-site inspection methods.