Data content mining: Extracting and cataloging content-based metadata from satellite images (remote sensing).

摘要

Data are not easily accessed without metadata within the National Aeronautics and Space Administration's (NASA) Earth Observing System Data and Information System (EOSDIS). Metadata is information about the inventory of data products managed and distributed by EOSDIS archive centers. Users interact with this information through an archive's catalog database to search for and locate data products used for research. Catalog metadata, however, are limited to high-level descriptions of the data product, not the content of the data. When EOSDIS archives begin accepting gigabytes of new data daily, innovative ways will be required for users to efficiently and effectively search these huge inventories for relevant data products. A promising solution to this problem is based on data mining to extract useful information about the content of data and augment catalog metadata with it. Development of a research prototype demonstrates how spectral histograms and inductive analyses can be used to establish new content-based search capabilities that address this problem.