Change detection is an important task in Earth observation, which monitors changes in land cover, land use, and environmental conditions over time. Employing bi-temporal images from remote sensing platforms like satellites, change detection is essential for managing both natural and human-induced transformations. Change detection employs both supervised and unsupervised methods at pixel or object-based scales. Recent trends incorporate deep learning techniques, improving the effectiveness and accuracy of change detection algorithms. Over the past decade, the increase in available remote sensing sensors has led to diverse data sources, including a number of multi-spectral, Synthetic Aperture Radar (SAR), and hyperspectral sensors at different spatial and temporal resolutions. Mentionworthy advancements involve the integration of multi-temporal data from multiple sensors, addressing the trade-off between spatial resolution and temporal frequency. Among other benefits, multi-sensor change detection proves advantageous in regions with frequent cloud cover. Additionally, multi-sensor data offer several other benefits in change detection, e.g., reduced uncertainty in results if used as complementary data sources. However, some unique challenges arise in processing and comparing data from different sensors and many research works have been proposed dedicated to tackle these challenges. This chapter provides a discussion on the research in multi-sensor change detection, initially outlining traditional methods and then emphasizing deep learning approaches. Additionally, it includes a couple of case studies centered around this topic.

Multi-sensor deep learning for change detection

Saha, Sudipan;Bergamasco, Luca;Atanasova, Milena;Bovolo, Francesca
2025-01-01

Abstract

Change detection is an important task in Earth observation, which monitors changes in land cover, land use, and environmental conditions over time. Employing bi-temporal images from remote sensing platforms like satellites, change detection is essential for managing both natural and human-induced transformations. Change detection employs both supervised and unsupervised methods at pixel or object-based scales. Recent trends incorporate deep learning techniques, improving the effectiveness and accuracy of change detection algorithms. Over the past decade, the increase in available remote sensing sensors has led to diverse data sources, including a number of multi-spectral, Synthetic Aperture Radar (SAR), and hyperspectral sensors at different spatial and temporal resolutions. Mentionworthy advancements involve the integration of multi-temporal data from multiple sensors, addressing the trade-off between spatial resolution and temporal frequency. Among other benefits, multi-sensor change detection proves advantageous in regions with frequent cloud cover. Additionally, multi-sensor data offer several other benefits in change detection, e.g., reduced uncertainty in results if used as complementary data sources. However, some unique challenges arise in processing and comparing data from different sensors and many research works have been proposed dedicated to tackle these challenges. This chapter provides a discussion on the research in multi-sensor change detection, initially outlining traditional methods and then emphasizing deep learning approaches. Additionally, it includes a couple of case studies centered around this topic.
2025
9780443264849
File in questo prodotto:
File Dimensione Formato  
sahasu12.pdf

solo utenti autorizzati

Tipologia: Documento in Pre-print
Licenza: NON PUBBLICO - Accesso privato/ristretto
Dimensione 2.75 MB
Formato Adobe PDF
2.75 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11582/355027
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
social impact