Compression Ratio as Indicator of Scientist Preference for Rover Images

Remote exploration spacecraft can generally cap-ture far more images than they can transmit over their bandwidth-constrained downlinks to Earth [1]. This sug-gests that onboard image analysis might improve the quality of returned science data: remote spacecraft gath-ering images at a high rate can select the most interesting data products for transmission [2]. Here we investigate statistical correlates of desirable image content to inform automatic image selection procedures. We focus on im-ages ’ compressed size, which functions as a rough mea-sure of the amount of “visual information ” present in the image. This study examines a 2003 survey on planetary scientists ’ preferences for surface rover images. Scien-tists ranked images according to their assessment of the images ’ science value, professing a wide range of ratio