This is the authors' abstract. We don't add key points for this paper.
The significant growth in satellite data has provided access to timely information of surface or airborne objects from almost anywhere on Earth. Recent interest has been garnered towards the remote Arctic region and coastal areas, particularly with the modernization of the North American Aerospace Defense Command (NORAD) to improve surveillance capabilities. The exploitation of large volumes of space-borne data for processing has added value for maritime surveillance. However, there exist gaps in how to effectively combine multi-satellite information and leverage multi-domain datasets. These include optical satellite imagery, optical satellite videos, synthetic aperture radar (SAR) imagery, as well as navigation information. We have developed a processing system for near real-time vessel detection, tracking and identification in space-based maritime surveillance applications. Our system architecture integrates machine learning and information fusion for situation assessment, and adopts ideas in generative machine learning to improve the vessel detection performance. We also fuse with data from automatic identification system (AIS) and tracks to provide information about the type of vessels. Modules developed for detection and tracking is verified with benchmark datasets and integrated into the system architecture. The system was successfully validated on real-world maritime surveillance scenarios including naval exercises from Rim of the Pacific 2020 and from various Skysat satellite videos.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Ma et al. (2024) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: