Open AccessA Method for Object Classification in Aerial/Satellite Images with Incorporating Geospatial Information
Author(s) -
Mykhailo Popov,
Мaksym V. Тopolnytskyi,
Valentyn Pylypchuk
Publication year - 2022
Publication title -
advances in military technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.137
H-Index - 11
eISSN - 2533-4123
pISSN - 1802-2308
DOI - 10.3849/aimt.1484
Subject(s) - geospatial analysis , multispectral image , computer science , geodetic datum , satellite , land cover , artificial intelligence , camouflage , remote sensing , object (grammar) , process (computing) , computer vision , data mining , geography , cartography , land use , engineering , civil engineering , aerospace engineering , operating system
Aerial and satellite multispectral images are important source of intelligence information. However, the object classification accuracy in those images for reasons such as camouflage, use of decoys, and others often turns out to be insufficient. The objective of the study is to develop a method for computer-aided analysis of aerial and satellite multispectral images, which allows improving classification accuracy. This objective is achieved by incorporating geospatial information (topographic, geodetic, about land cover types) into the classification process. As a mathematical basis of the method is used subjective logic of A. Jøsang. The effectiveness of the proposed method has been demonstrated by computer modeling using ArcGIS ModelBuilder tools.