Compression and Sensing in Earth Observation and Beyond

The field of data compression and sensing is moving towards innovative solutions that leverage tensor-based methods, analog compressed sensing, and learned codecs. Recent developments have shown that standard SVD-based compression approaches can be effective for spatiotemporal data, while approximate proximal operators can be realized using electric analog circuits. Furthermore, learned codecs tailored to Earth observation have achieved significant compression ratios, outperforming classical codecs. Noteworthy papers include TerraCodec, which introduces a family of learned codecs for Earth observation, and A Dimension-Keeping Semi-Tensor Product Framework for Compressed Sensing, which proposes a novel method for compressed sensing that leverages intra-group correlations. These advancements have the potential to enable more efficient storage and transmission of large datasets, and improve the reconstruction performance of sparse signals.

Sources

Tensor-based compression of the sea temperature data

Approximate Proximal Operators for Analog Compressed Sensing Using PN-junction Diode

TerraCodec: Compressing Earth Observations

A Dimension-Keeping Semi-Tensor Product Framework for Compressed Sensing

A 0.62 {\mu}W/sensor 82 fps Time-to-Digital Impedance Measurement IC with Unified Excitation/Readout Front-end for Large-Scale Piezo-Resistive Sensor Array

Built with on top of