Compressive sensing instrumental concepts for space applications

The need of high-resolution Earth Observation (EO) images for scientific and commercial exploitation has led to the generation of an increasing amount of data with a material impact on the resources needed to handle data on board of satellites. In this respect, Compressive Sensing (CS) can offer interesting features in terms of native compression, onboard processing and instrumental architecture. In CS instruments the data are acquired natively compressed by leveraging on the concept of sparsity, while on-board processing is offered at low computational cost by information extraction directly from CS data. In addition, instrument’s architecture can enjoy super-resolution capabilities that ensure a higher number of pixels in the reconstructed image with respect to that natively provided by the detector. In this paper, we present the working principle and main features of a CS demonstrator of a super-resolved instrument for EO applications with ten channels in the visible and two channels in the medium infrared. Besides the feature of merging in a single step the acquisition and compression phases of the image generation, its architecture allows to reach a superresolution factor of at least 4x4 in the images reconstructed at the end of process. The outcome of the research can open the way to the development of a novel class of EO instruments with improved Ground Sampling Distance (GSD) - with respect to that one provided natively by the number of sensing elements of the detector - and impact EO applications thanks to native compression, on-board processing capabilities and increased GSD.