Building classification in Yangon City, Myanmar using Stereo GeoEye images, Landsat image and night-time light data

Yangon, the former capital of Myanmar, is the biggest city in the country with more than five million people and it is also the major country's economic areas. These areas are complex with three classes of residential, commercial and industrial buildings. Understanding building uses in the city with the information of the locations and the quantitative measurement is very important to support urban management and development with various aspects. This research proposed a methodology to classify types of buildings with three classes in Yangon, Myanmar by using remotely sensed data. In this research, stereo GeoEye images, multi-spectral Landsat image and night-time light (NTL) image from Visible Infrared Imaging Radiometer Suite (VIIRS) were employed to extract types of buildings. The Stereo GeoEye images were used to obtain the heights of buildings, the Landsat image was classified to provide land cover areas, and NTL image was applied to separate NTL activities. By using the hierarchy classification with (1) the heights of buildings, (2) land cover areas and (3) NTL consumptions, the buildings were classified into three classes with (1) residential, (2) commercial, and (3) industrial buildings. In the experiments, the estimated building type map by our proposed method was compared with a land use map and surveying building data. The comparing results indicated that our methodology classified types of buildings in efficiency with the accuracy of 76% and the Kappa coefficient of 0.58.