Abstract
Accurate built-up area extraction is one of the most critical issues in land-cover classification. In previous studies, various techniques have been developed for built-up area extraction using Landsat images. However, the efficiency of these techniques under different technical and geographical conditions, especially for bare and sandy areas, is not optimal. One of the main challenges of built-up area extraction techniques is to determine an optimum and stable threshold with the highest possible accuracy. In many of these techniques, the optimum threshold value fluctuates substantially in different parts of the image scene. The purpose of this study is to provide a new index to improve built-up area extraction with a stable optimum threshold for different environments. In this study, the developed Automated Built-up Extraction Index (ABEI) is presented to improve the classification accuracy in areas containing bare and sandy surfaces. To develop and evaluate the accuracy of the new method for built-up area extraction with Landsat 8 OLI reflective bands, five test sites located in the Iranian cities (Babol, Naqadeh, Kashmar, Bam and Masjed Soleyman), eleven European cities (Athens, Brussels, Bucharest, Budapest, Ciechanow, Hamburg, Lyon, Madrid, Riga, Rome and Porto) and high resolution layer imperviousness (HRLI) data were used. Each site has varying environmental and complex surface coverage conditions. To determine the optimal weights for each of the Landsat 8 OLI reflective bands, the pure pixel sets for different classes and the improved gravitational search algorithm (IGSA) optimization were used. The Kappa coefficient and overall error were calculated to evaluate the accuracy of the built-up extraction map. Additionally, the ABEI performance was compared with the urban index (UI) and normalized difference built-up index (NDBI) performances. In each of the five test sites and eleven cities, the extraction accuracy of the built-up areas using the ABEI was higher than that using the UI, and NDBI (P-value of 0.01). The relative standard deviations of the optimal threshold values for the ABEI and UI were 27 and 155% (at five test sites) and were 16 and 37% (at eleven European cities), respectively, which indicates the stability of the ABEI threshold value when the location and environmental conditions change. The results of this study demonstrated that the ABEI can be used to extract built-up areas from other land covers. This index is effective even in bare soil and sandy areas, where other indices experience major challenges.
Highlights
Urban areas generally cover a small portion of land worldwide
4.1.TAhBeEaIim of the Automated Built-up Extraction Index (ABEI) is to create the greatest similarity between pixel values within the built-up class aTnhdetahiemgorefattheestAdBifEfIerisentocecrineapteixtehlevgarleuaetsesbtestwimeielnartihtye bbeutiwlte-uenp pcliaxsesl vanaldueosthwerithclianssthese.bInuitlht-iusp stculadsys, aunsdintgheIGgrSeAateospttdimiffizearetinocne, itnhepioxbeljevcatilvueesfubnetcwtioenenatnhde bpuuirlet-uppixcellasssofanddiffoetrheenrt csluasrsfaecse
The main objective of this study was to develop an automatic technique that improves the accuracy of built-up extraction by increasing spectral separability between built-up and non-built-up surfaces, in regions with bare land and sandy soils, which are often major causes of low classification accuracy
Summary
Urban areas generally cover a small portion of land worldwide. the high population density and intensity of utilizing resources relative to their surroundings turn these areas into highly important land areas [1,2]. Understanding the spatial distribution and growth pattern of urban built-up areas is vital for urban planning, natural resource management, climate change monitoring, etc., and the ecological, economic and social effects of built-up lands have made the mapping of built-up areas very important [2,3,4,5,6]. Remote sensing technology provides an integrated, comprehensive view of urban areas that is difficult to achieve through surveying. Another advantage of using remote sensing data for urban studies is the availability of archives of past periods that can be used to review, monitor, and prepare a distribution map of built-up areas over time [13,14]. Satellite imagery capabilities have been used for studies on urban expansion [15], effective factors on urban development [16], and the negative effects of urban growth [17]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
