Pregled bibliografske jedinice broj: 1221080
A Comparative Assessment Between Pixel and Geoobject based Classification Approaches for Land Use/Land Cover Mapping Using VHR Multispectral Images
A Comparative Assessment Between Pixel and Geoobject based Classification Approaches for Land Use/Land Cover Mapping Using VHR Multispectral Images // 8th International Scientific Conference GEOBALCANICA 2022.
Beograd, Srbija; online, 2022. (poster, međunarodna recenzija, sažetak, znanstveni)
CROSBI ID: 1221080 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
A Comparative Assessment Between Pixel and
Geoobject based Classification Approaches for Land
Use/Land Cover Mapping Using VHR Multispectral
Images
Autori
Šiljeg, Ante ; Mulahusić, Admir ; Hadžić, Nudžejma ; Domazetović, Fran ; Panđa, Lovre
Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni
Izvornik
8th International Scientific Conference GEOBALCANICA 2022.
/ - , 2022
Skup
8th International Scientific Conference GEOBALCANICA 2022.
Mjesto i datum
Beograd, Srbija; online, 09.05.2022. - 10.05.2022
Vrsta sudjelovanja
Poster
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
LULC ; image classification ; Random Trees ; Maximum Likelihood ; Support Vector Machine ; Metković
Sažetak
LULC mapping is essential in environmental research, urban planning, vegetation studies, and water and soil management. The main goal of this paper was to compare GEOBIA and PB classification algorithms. The best model is ML (GEOBIA) with an overall accuracy of 59.81%. To increase the accuracy on VHR models it is necessary to adjust the segmentation parameters separately for each class and to collect larger number of test samples and LULC classes.
Izvorni jezik
Engleski
Znanstvena područja
Interdisciplinarne prirodne znanosti, Geografija
POVEZANOST RADA
Ustanove:
Sveučilište u Zadru