SUM Feature selection and preparation for fusion (Part: Radiometer characterization)

Abstract

The main objectives of this work-package include: (i) The adaptive selection of features, in order to fuse only relevant data from the individual sensors hence reduce the number of alerts and accelerate the fusion process, and (ii) the preparation of the outputs for fusion. The former objective relies on the outcome of features that originate from sensor or the vehicle as well as prior information resulting from data collected by UAV and satellite sensors. Selecting appropriate features and their properties necessitates defining scenarios that describe the considered threats. The beginning of this document concerns this topic. Furthermore, this document describes in general terms the image/video processing techniques adopted for extracting features from ground vehicles sensors and those emanating from UAV/Satellite sources. Hence it outlines the data-flow starting from raw image/video data to an actual feature, which in turn can be characterized to describe one or more threats. Finally, the preparation for fusion concludes with listing the considered threats in terms of sensor input, prior knowledge, environment conditions, and detection accuracy.