Preliminary communication
https://doi.org/10.21861/HGG.2014.76.02.02
Land Cover Classification of Urban and Peri-urban Areas Using Object-oriented Analysis of Multispectral Imagery
Luka Valožić
orcid.org/0000-0001-5696-0367
; Sveučilište u Zagrebu, Prirodoslovno-matematički fakultet, Geografski odsjek, Zagreb, Hrvatska
Abstract
Supervised classification of remote sensing imagery implies usage of a priori knowledge of analysed area of interest for proper selection of land cover classes’ samples. Spectral and geometric properties of the satellite image were taken into account by parameters of the multiresolution segmentation algorithm that has been used to divide the selected subset of the scene into homogeneous groups of pixels. Such image segments were used as land cover samples for supervised classification. Accuracy assessment was performed by means of error matrix based on stratified random samples.
Keywords
land cover; satellite images; method of supervised classification; image segmentation; error matrix; V-I-S model
Hrčak ID:
135008
URI
Publication date:
23.2.2015.
Visits: 2.377 *