1. Summary

The main target of my research is visual data like images, and especially I focus on satellite images, or remote sensing images, from a viewpoint of realizing a unified infrastructure for the archive, retrieval, data mining, and presentation of a large-scale scientific data collection. Along with this goal, we also keep international research collaboration between Asian Institute of Technology in Thailand to realize near-realtime satellite data distribution in an international scale. => Brief Introduction (From NII News).

2. History and Usage

3. Demonstrations

4. Satellite Image Processing

The principal goal of this research is to study the paradigm of informatics-based approaches for the effective uses of satellite images by both scientists and people in general in order to improve the accessibility to relevant satellite data which have much potential to solve earth environmental issues. Toward this goal, we will build and maintain huge-scale scientific databases, in which the meteorological satellite images are the specific target, and build a theory for scientific databases through abstracting our experience with the meteorological image database platform.

To be more specific, our main target among meteorological satellites is what is called a geostationary metorological satellite, which is positioned above the equator and moves along with the earth's rotation. This type of satellite can repeat the observation from the Arctic to the Antarctic every one hour, and suitable for monitoring the meteorological phenomena from a global scale to a regional scale. Because of the power of the image data, this satellite data is indispensable not only as a basic dataset for weather forecast, but also as an attractive material for weather forecasting programs on TV and media.

(Note) The "GMS" (aka Himawari) series were this type of satellite in the Pacific area, but after May 22, 2003, the Japanese GMS-5 satellite was switched over to American GOES-9 satellite. Related information can be found in the following page: From GMS-5 to GOES-9.

Thus observation data were fully used for understanding the current state of the atmosphere; however, at the same time, it is wasteful if we just discard those data for temporary uses only and not for scientific uses in terms of mining some knowledge from the data collection. It is therefore an important challenge to establish a principle for collecting, archiving, searching and organizing meteorological datasets in a very large scale. Moreover, we will also refer to special problems with environmental satellite data, such as how to exchange huge amount of satellite data between data centers distributed over the Internet, or how to present the contents of the data which may not be easy for non-experts to understand. Specific topics are as follows.

1. Content-based Image Retrieval Technology for large-scale satellite data collections

Typical search engines for earth-monitoring satellites are based on the metadata of satellite observations, such as observation time or satellite name. However, more powerful search engines can be realized by content-based image retrieval technology, which is used for automatically extracting important image features and make an index for image contents.

2. Networking Technology for distributed satellite data archives

Typically satellite data receiving stations are distributed over the earth, and they archive their own unique data. So the networking technology for connecting those distributed centers and build a unified data collection platform for searching and collecting necessary data. Similar ideas are now called "Data Grid" and are attracting more and more attention.

3. Visualization Technology for the presentation of various information contained in the satellite data in an easily understandable way

Satellite data are basically an array of numbers, and you cannot fully understand the contents by just looking through those numbers. It is therefore really important to extract relevant information from raw data, and present such relevant information in an easily understandable way. This kind of technology is generally called as "visualization", but recently, a related keyword "visual data mining" is also gaining popularity.

4. Marriage with Geoinformatics

Remote sensing and geoinformatics can "give and take" each other in terms of the following points, so the relationship between those two fields will deepen more in the future.

  1. Remote sensing data are geographical data in the sense that every pixel corresponds to a point (or a region) in some geographical coordinates. If we treat remote sensing data not as a mere image data but as a geographical data, we can incorporate prior information associated with the geographical location, and use that information for image analysis, for example. Thus we should treat remote sensing data as a geographical data, or spatio-temporal data including the temporal coordinate.
  2. On the other hand, from the viewpoint of geoinformatics, the rapid and comprehensive data collection in a global scale using remote sensing is an indispensable technology for building a GIS (Geographic Information System) that dynamically adapts to a rapid change of geographical information, such as in the case of the occurrence of disasters. In addition, the technology for collecting and archving earth environment information in a comprehensive and uniform manner will become the key technology for a global-scale geographic information infrastructure such as "Digital Earth."

5. Earth Observed from Satellites

  1. Geostationary Meteorological Satellites: Himawari, GOES, GMS and MTSAT
  2. Earth in the Daytime from Space: The Blue Marble Data by MODIS Satellites
  3. Earth in the Night from Space: Nighttime Lights of the World Data by DMSP Satellites
  4. Earth in the Night from Space: Nighttime Lights Time Series by DMSP Satellites (1992-2012 : Google Maps Version)
  5. Citadel of Bam: Very High Resolution Satellite Images Before and After the Earthquake
  6. Sea ice from meteorological satellites
  7. Volcano eruption from meteorological satellites
  8. Solar eclipse from meteorological satellites

6. References (Complete List)

  1. Asanobu KITAMOTO, "Remote Sensing: From Image Processing to Spatio-temporal Processing", Technical Report of IEICE (Institute of Electronics, Information, and Communication Engineers), Vol. PRMU2002-255, pp. 73-80, 2003-03 (in Japanese) [ Abstract ] [ PDF ]
  2. Asanobu KITAMOTO, Kinji ONO, "The Collection of Typhoon Data and the Construction of Typhoon Image Databases Under International Research Collaboration between Japan and Thailand", NII Journal, No. 2, pp. 15-26, 2001-03 [ Abstract ] [ PDF ]
  3. Asanobu KITAMOTO, "Multiresolution Cache Management for Distributed Satellite Image Database Using NACSIS-Thai International Link", Proceedings of the 6th International Workshop on Academic Information Networks and Systems (WAINS), pp. 243-250, 2000-03 [ Abstract ] [ PDF ]
  4. Asanobu KITAMOTO, "Toward Content-Based Satellite Image Database Systems over the Network", Proceedings of the 5th International Workshop on Academic Information Networks and Systems (WAINS), pp. 31-38, 1998-12 [ Abstract ] [ PDF ]
  5. Asanobu KITAMOTO, Mikio TAKAGI, "3D Visualization of Meteorological Satellite Imagery Using Volume Rendering", Proceedings of the Annual Convention, The Institute of Television Engineers of Japan, pp. 114-115, 1996-07 (in Japanese) [ Abstract ] [ PDF ]

7. Related Links