COMPARISON OF PIXEL AND OBJECT-BASED APPROACH FOR BUILDING ROOF MAPPING USING MULTI-SPECTRAL AND PAN-SHARPENED WORLDVIEW-2 IMAGES

The detailed scale mapping on building roofs using remote sensing imagery for information extraction. The geographic Based-Image Analysis (GEOBIA) classification method is known as an alternative pixel-based classification method such as Random Forest. This study aims to compare the effectiveness of pixel-based and object-based classification methods in obtaining information on building rooftop classes mapped using images with different spatial resolutions. The research results indicate that: 1) GEOBIA classification yields higher overall accuracy for multispectral and pan-sharpened images than the Random Forest method. 2) WorldView-2 Multispectral imagery produces lower accuracy compared to WorldView-2 Pan-sharpened imagery. © 2023 ACRS. All Rights Reserved.