|
A number of web-accessible databases, including medical, military or other image data, offer universities and
other users the ability to teach or research new Image Processing techniques on relevant and well-documented
data. However, NASA images have traditionally been difficult for researchers to find, are often only available in
hard-to-use formats, and do not always provide sufficient context and background for a non-NASA Scientist
user to understand their content. The new IMAGESEER (IMAGEs for Science, Education, Experimentation and
Research) database seeks to address these issues. Through a graphically-rich web site for browsing and
downloading all of the selected datasets, benchmarks, and tutorials, IMAGESEER provides a widely accessible
database of NASA-centric, easy to read, image data for teaching or validating new Image Processing algorithms.
As such, IMAGESEER fosters collaboration between NASA and research organizations while simultaneously
encouraging development of new and enhanced Image Processing algorithms. The first prototype includes a
representative sampling of NASA multispectral and hyperspectral images from several Earth Science
instruments, along with a few small tutorials. Image processing techniques are currently represented with cloud
detection, image registration, and map cover/classification. For each technique, corresponding data are selected
from four different geographic regions, i.e., mountains, urban, water coastal, and agriculture areas. Satellite
images have been collected from several instruments - Landsat-5 and -7 Thematic Mappers, Earth Observing -1
(EO-1) Advanced Land Imager (ALI) and Hyperion, and the Moderate Resolution Imaging Spectroradiometer
(MODIS). After geo-registration, these images are available in simple common formats such as GeoTIFF and
raw formats, along with associated benchmark data.
|
