Mining in Image & Raster Databases
Magellan project (Fayyad et al.’96, JPL).
- Identify volcanos on Venus surface
- ᡖ,000 high resolution images
- Resolution accuracy: 80%
- 3 steps: data focusing, feature extraction, and classification learning
POSSII project (Palomar Obervatory Sky Survey II, )
- 2 x 109 stellar objects (galaxies, stars, etc.) classified
- Resolution:one magnitude better than in previous studies
- Classification accuracy: no normalization 75%, with normalization 94%, and compared with neural networks.
QuakeFinder (Stolorz et al’96): Find earth quakes from space.
- using statistics, massive parallelism, and global optimization