@article{oai:nipr.repo.nii.ac.jp:00004212,
 author = {Kurozumi, Yoshisuke and Maeda, Sawako and Tamenaga, Hiroki and Miyaoka, Hiroshi and Kadokura, Akira and Yukimatsu, A.Sessai},
 journal = {Proceedings of the NIPR Symposium on Upper Atmosphere Physics},
 month = {Jan},
 note = {P(論文), Image Database Systems usually have some keywords selected by human decision. In recent years, it has been an interesting problem to develop a method by which shapes, colors and contents for a specific image from image files are recognized automatically through image processing. Furthermore, it becomes important to determine properly the keywords representing impression of images, that is, 'KANSEI' words. We try to construct a database and data search system of the DMSP satellite auroral image files. At first, an index file including information such as file name, size, date and geometric feature of images is created through image processing. The impressional items for auroral shapes such as brightness, position, and impression are derived by a coordinate conversion method so called the 'KANSEI' space method. The system contains about 100 image files and the index file. A specific image data is searched with the index file and the conversion method of 'KANSEI' words.},
 pages = {117--125},
 title = {Database and data search system using 'KANSEI' words for aurora image files of DMSP satellites},
 volume = {10},
 year = {1997}
}