Key points are not available for this paper at this time.
Imaging sensors are inherently high bandwidth devices, and applications which store image data often encounter disk or memory limits. Commonly, upon reaching such a limit, storage systems will cease sampling or overwrite existing data in an oldest-first fashion. For most applications, neither approach is optimal. We introduce a flexible, policy-based, on-line algorithm for maximizing the application-specific value of data retained on a read/write/erase storage medium such as a hard disk or flash card. Our algorithm supports both selective discard of image frames, and selective lossless and lossy transformations which may retain information valuable to an application while reducing data storage requirements. We demonstrate that our multi-fidelity storage system can fairly manage a limited data storage medium on behalf of one or more image storage applications.
Pillai et al. (Fri,) studied this question.