Sentinel-2 műholdfelvételek felügyelet nélküli osztályozása
An effective approach for monitoring global change is feasible via the use of satellite images. These images provide important information that aids in observing various applications such as change detection, monitoring disasters, natural hazards, or land cover classification. Remote sensing is a key technique used to obtain information related to the Earth’s resources and environment.
The Sentinel-2 satellites provide global coverage of land surfaces with a 5-day revisit time at the equator. These multi-spectral instruments mainly applied in agriculture, such as crop monitoring and management, vegetation and forest monitoring, tracking land cover change for environmental analysis, observation of coastal zones, inland water and glacier monitoring, ice extent, and snow cover mapping.
In this study, High-resolution satellite images were obtained from the Sentinel-2 Copernicus Open Access Hub, and an unsupervised classification algorithm was applied to the image. The K-means clustering algorithm is used particularly as it delivers training results quickly where images are unlabeled. The aim of this study to investigate sequentially (time series) satellite image classifications of Sentinel-2, after running the clustering algorithm on these images relevant conclusions are being made: like the cluster of vegetation changes, calculating surface differences, and changes in land use over time.
EMMANUEL GODWIN BANDAWA
Építőmérnöki szak (műszaki alapdiploma BSc szint)
Dr. Wirth Ervin
adjunktus, Fotogrammetria és Térinformatika Tanszék
Dr. Lovas Tamás
egyetemi docens, Fotogrammetria és Térinformatika Tanszék