Systems and methods for expert guided rule based identification of relevant planetary images for downlinking over limited bandwidth

A system, comprising:

• A processor operable to execute instructions to:
• emulate expert-like identification of relevant datasets by focusing on predetermined geological features, thereby maximizing the return of high value data over limited bandwidth, wherein to emulate, the processor:
• accesses a dataset, wherein the dataset comprises a plurality of feature instances and wherein each of the plurality of feature instances are associated with a feature class of a plurality of feature classes;
• updates a set of class metrics based on each of the plurality of feature instances in the dataset, wherein each of the set of class metrics are associated with expected properties belonging to each of the plurality of feature classes;
• accesses a plurality of new feature instances, wherein each of the plurality of new feature instances are associated with one of the plurality of feature classes;
• evaluates a set of novelty metrics for each of the plurality of new feature instances based on the set of class metrics and the plurality of new feature instances, wherein the set of novelty metrics are associated with how each of the plurality of new feature instances relate to expected properties belonging to a given associated feature class;
• combines the set of novelty metrics as evaluated for each of the plurality of new feature instances to assign an overall relevance score to each of the plurality of new feature instances, wherein a given new feature instance of the plurality of new feature instances having a high overall relevance score indicates a high priority associated with the given new feature instance;
• and
• updates the dataset to include each of the plurality of new feature instances.