%0 Conference Proceedings %B 32nd Asian Conference on Remote Sensing %D 2011 %T High-Resolution Urban Area Map for 3372 Cities of the World %A Miyazaki, Hiroyuki %A Itabashi, Koichiro %A Shao, Xiaowei %A Iwao, Koki %A Shibasaki, Ryosuke %K ASTER %K automated image selection %K automated urban area mapping %K global urban area map %K Learning with Local and Global Consistency %K logistic regression %K multi-source classification %X We developed an automated method of image selection and urban area mapping for developing high-resolution global urban area maps. The method was successfully implemented and applied to 3372 cities of more than 0.1 million people of the world. As the result, the algorithm of image selection determined 11802 scenes of ASTER/VNIR satellite images and yielded good combinations for more than 60% of the cities. For the merged satellite images with the determined combinations, we applied the automated method of urban area mapping in high resolution. The method was consist of semi-supervised classifications by a machine learning method, called Learning with Local and Global Consistency (LLGC), and integrating the LLGC-derived maps and existing maps by logistic regression. As the result, we acquired urban area maps of 15-m resolution originated from ASTER/VNIR images, which is much finer than 500-m resolution of existing urban area maps. The method had still much space to be improved, especially in avoiding cloud contaminations. However, the method would contribute to complete high-resolution urban area maps of the world and realizing Global Earth Observation Systems of System (GEOSS). %B 32nd Asian Conference on Remote Sensing %C Taipei %P PS–3 %G eng