IMAGESEER is a new Web portal that brings easy access to NASA image data for non-NASA researchers, educators, and students. The IMAGESEER Web site and database are specifically designed to be utilized by the university community, to enable teaching image processing (IP) techniques on NASA data, as well as to provide reference benchmark data to validate new IP algorithms. Along with the data and a Web user interface front-end, basic knowledge of the application domains, benchmark information, and specific NASA IP challenges (or case studies) are provided.
Working with project scientists and engineers, four types of IP techniques have been identified as corresponding to Earth Science needs; these are gap filling/in-painting, cloud detection, image registration, and map cover/classification. For each of these challenges, corresponding data were selected from four different geographic regions: mountains (Colorado), urban (Los Angeles), water coastal area (Chesapeake Bay), and agriculture (Illinois). Satellite images have been collected for these areas from several satellite instruments, then georegistered, and finally converted to common image formats (GeoTIFF and raw). Along with the original data, associated benchmarks (or validation data) have been acquired or generated, including cloud cover masks and assessments, georegistered scenes, and classification maps from the National Land Cover Data (NLCD) database gathered in 1992 and 2001 by the Multi- Resolution Land Characteristics Consortium (MRLC).
IMAGESEER provides a modern and graphically-rich Web site (https://imageseer.nasa.gov) for easily browsing and downloading all of the selected datasets, benchmarks, and tutorials. Using a paradigm common on commercial Web sites, users can restrict their searches by selectively filtering by data source (project, mission, and instrument), region of interest, desired image processing technique, and time period. By deliberately focusing on only a subset of NASA data, continuously emphasizing ease-of-use, providing common file formats, and supplying the “answers” as well as the questions to NASA IP challenges, IMAGESEER provides an easily navigable and usable Web site for non-NASA researchers, educators, and students. On the backend, automated Python scripts convert the NASA data, generate thumbnails and benchmarks, and populate the IMAGESEER database. The database and the IMAGESEER Web site were developed using a MySQL database and Hyper Text PreProcessor (PHP).
IMAGESEER currently focuses on Earth Science data, but is designed to be straightforwardly extended to planetary and exploration data, and in fact, planetary-specific challenges, such as automated crater counting and boulder counting, have already been identified. IMAGESEER is ideal for helping educators and students learn IP techniques needed for and with actual NASA data and for helping researchers develop new algorithms or adapt existing algorithms to NASA data and challenges. It provides a focused set of NASA-centric data for education and research, hopefully engendering further interest in NASA careers and research.
This work was performed by Jacqueline Le Moigne, Thomas Grubb, and Barbara Milner for Goddard Space Flight Center. GSC-15967-1