AI4EO: from physics guided paradigms to quantum machine learning

Abstract

Earth Observation (EO) Data Intelligence is addressing the entire value chain: data processing to extract information, the information analysis to gather knowledge, and knowledge transformation in value. EO technologies have immensely evolved the state of the art sensors deliver a broad variety of images, and have made considerable progress in spatial and radiometric resolution, target acquisition strategies, imaging modes, geographical coverage and data rates. Generally imaging sensors generate an isomorphic representation of the observed scene. This is not the case for EO, the observations are a doppelgänger of the scattered field, an indirect signature of the imaged object. EO images are instrument records, i.e. in addition to the spatial information, they are sensing physical parameters, and they are mainly sensing outside of the visual spectrum. This positions the load of EO image understanding, and the outmost challenge of Big EO Data Science, as new and particular challenge of Machine Learning (ML) and Artificial Intelligence (AI). The presentation introduces specific solutions for the EO Data Intelligence, as methods for physically meaningful features extraction to enable high accuracy characterization of any structure in large volumes of EO images. The theoretical background is introduced, discussing the advancement of the paradigms from Bayesian inference, machine learning, and evolving to the methods of Deep Learning and Quantum Machine Learning. The applications are demonstrated for: alleviation of atmospheric effects and retrieval of Sentinel 2 data, enhancing the opportunistic bi-static images with Sentinel 1, explainable data mining and discovery of physical scattering properties for SAR observations, and natural embedding of the PolSAR Stokes parameters in a gate-based quantum computer.