سال انتشار: ۱۳۹۰

محل انتشار: پنجمین کنفرانس بین المللی پیشرفتهای علوم و تکنولوژی

تعداد صفحات: ۸

نویسنده(ها):

m ghanea – Dept. of Surveying Eng., Faculty of Eng., University of Isfahan, Isfahan
p moallem – Dept. of Electrical Eng., Faculty of Eng., University of Isfahan, Isfahan
m momeni – Dept. of Surveying Eng., Faculty of Eng., University of Isfahan, Isfahan

چکیده:

An inclusive accessibility of very high resolution (VHR) satellite images has generated more interest in automatic extraction of buildings for some practical applications such as updating geographic information system (GIS) database, land management, cadastre, and 3D city modeling. Extraction of buildings especially in a dense urban area containing many different and connected parts is an intricate problem due to a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as a variety of difficulties such as shadows and similar spectra. In order to face this problem properly, we present an algorithm for extraction and improvement of buildings. The process consists of four steps: (1) separate building and non-building pixels, (2) improve the building layer, and (3) segment pixels that belong to the building layer. The proposed algorithm is evaluated for a case study in Tehran, Islamic Republic of Iran using a pan sharpened multispectral GeoEye satellite image. The experiments show that the algorithm extracts 78.3 % of buildings with a quality percentage 59.7 % in a dense urban area