An overview of compression strategies for scientific imaging
by
OHSA/B17
Abstract:
High-speed x-ray time-resolved tomography produces data streams of several terabytes per hour, pushing acquisition, transfer, and storage pipelines to their limits. This talk surveys compression strategies tailored for scientific imaging—from classical transform and predictive coding to video-based codecs and modern neural approaches. Emphasis is placed on preserving quantitative detail for accurate 3D/4D reconstructions while achieving real-time throughput. I will highlight current capabilities, trade-offs between efficiency and fidelity, and future directions for integrating compression directly into tomography acquisition pipelines. I also describe the goals of our SDSC project SDATE: Smart Data Acquisition for Tomoscopy Experiments and its relation to compression pipelines.
Remote participation: TEAMS
The Laboratory for Simulation and Modeling
SDSC Hub at PSI