Object-oriented classification for land cover of North Thang Long Industrial area using Worldview-2 data

Abstract

Land cover/land use classification using high spatial resolution remote sensing data has the biggest challenge is how to distinguish object classes from different spectral values based on structures, shapes, and spatial elements. This paper focuses on the object-oriented classification method to extract artificial surface at industrial area by Worldview-2 data with a spatial resolution of 1.8 m. Extraction of 05 types of land cover/land use using object-oriented classification method based on reflectance spectral characteristics, shape index, location of objects, brightness, NDVI index, and density objects are archive efficiency to the quality of classification results. The overall accuracy of classification result for land cover/land use of Thang Long industrial area is about 0.85 and Kappa index is about 0.81.