Real-Time Detection Method of Persistent Objects in Radar Imagery with Deep Learning

• Autores: Martinez R.G., Vera J.M., Arrese C.C.
• Formato: artículo
• Fecha de Publicación: 2020
• Institución: Consejo Nacional de Ciencia Tecnología e Innovación
• Repositorio: CONCYTEC-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorio.concytec.gob.pe:20.500.12390/2474
• Enlace del recurso: https://hdl.handle.net/20.500.12390/2474; https://doi.org/10.1109/EIRCON51178.2020.9254021
• Nivel de acceso: acceso abierto
• Materia: Radar images; Faster R-CNN; Persistent objects; http://purl.org/pe-repo/ocde/ford#2.02.03

Persistent object detection in radar imagery becomes harder if the results are expected before the next image arrives to the digitizer card. This requires a clear commitment between the hit rate, the false contact rate and a time restriction in order to get real-Time results taking one full revolution of the radar as the basic unit. The conventional algorithms use CFAR techniques and obtain acceptable results, but with a high false contact rate, especially in near-shore radar imagery, which contain ground clutter portions of the images. This work presents the first results of the analysis to the solutions to this problem by applying Deep Learning. This research proposes the use of convolutional neural networks Faster R-CNN on radar imagery. The developed methods are applied using a methodology. The purpose of this research is to provide methods and techniques to improve the detection of persistent objects, thus having a positive impact in the maritime control and surveillance operations. © 2020 IEEE.