Automatic 2D Building Extraction Using High Resolution Image in Bangpli District, Samut Prakan Province, Thailand

摘要

Most of the recent work in Thailand, on building extraction, is based on 3D building extraction by manual digitization from aerial photo which are both time and labor consuming. High-resolution satellite imagery is expected to be widely used on urban planning mapping, cartography, land use application and others.Automatic and/or semi automatic building extraction is one of the alternative to develop method for building features extraction, with different rooftops, from high resolution satellite image in 1 sq.km. of urban prawn, with mixed area, which is approximately 40% of the study area is residential, commercial area while other 30% is plantation area, in a part of Bangpli District, Samut Prakan Province, Central of Thailand. The image was implemented on the supervised classification function of PCI Geometica and feature analyst extension of ArcGIS software. The building extraction results were compared with percentage of the result. In this study, the result of supervised classification methodology can extract approximately 32% of buildings image properly and the other can extract approximately 42%. Exactly, the both method used less time, so they showed the time reduction, hardware, software, and labor in feature extraction process, although they were not perfect result at the first output. For future work, if the methods were improved algorithm, the result should get better.