Comparison of spectral indices and object- based classification for built-up area extraction in different urban areas

Following the increase in population and the growth of urbanization in recent decades, the natural landscape is becoming a human landscape, and the urban open space has become a constructed area. The aim of this study was to investigate the performance of four widely used urban spectral indices in order to separate the areas built from other uses in two different cities and compare it with the results of object-oriented classification to extract the constructed areas. For this purpose, the normalized difference indices of the constructed areas, the urban index, the normalized index of the constructed areas and the extraction index of the constructed areas from the OLI satellite images of Landsat 8 satellite for the two cities of Tehran and Gorgan have been calculated. Autos automatic thresholding method has been used to separate the constructed and un constructed areas. The research findings showed that spectral indicators in different urban areas do not have the same performance and the accuracy of extraction of constructed areas by applying these indicators does not exceed the accuracy of object-oriented classification. The overall accuracy of the object-oriented classification in extracting the constructed areas in Tehran is 92% and in Gorgan is 91%. The best spectral index of the constructed areas for the city of Tehran was the extraction index of the constructed areas with an overall accuracy of 82% and for the city of Gorgan, the normal index of the constructed areas with a total accuracy of 86%