A Study on Closed Region Extraction for building extraction from Simulated High Resolution Satellite Imagery.

概要

論文の詳細を見る
Up to now, aerial photographs have almost been the unique data sources to be used for the purpose of urban spatial data acquisition. Recently, great attention has been paid to the emergence of high resolution satellite imagery (resolution around 1 meter) . Huge benefits may be reached if the so-called high resolution satellite data can be used as the data source for urban spatial data acquisition, especially for building related information collection. In this paper, the characteristics of high resolution satellite imagery are analyzed at first. Two practical building extraction methods are then proposed. One method aims to extract buildings with image features (edges and corners) extracted from high resolution satellite images, while the other tries to extract buildings with region growing algorithm. Experiments are carried out for testing the efficiencies of proposed two building extraction methods. Since high resolution satellite images are still not available at the moment, three different types of simulation images with different resolutions (0.5m, lm and 3m, respectively) made from aerial photographs are used in experiments.
社団法人日本写真測量学会の論文