Using Onboard Clustering to Summarize Remotely Sensed Imagery.

The article reports that Jet Propulsion Laboratory (JPL) has developed algorithms using onboard clustering to analyze science imagery onboard spacecraft. It notes that various approaches were examined including k-means clustering with the use of image features based on color, texture and temporal in comparison to the manual clustering of a field expert. The challenges influencing JPL in their design decisions for automatic image analysis include limited onboard processing in spaceflight applications and diversity in surface features.