BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.staging.osgeo.org//foss4g-it-2023//talk//URR3NB
BEGIN:VTIMEZONE
TZID:GMT
BEGIN:STANDARD
DTSTART:20001029T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:GMT
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T020000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:BST
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-foss4g-it-2023-URR3NB@talks.staging.osgeo.org
DTSTART;TZID=GMT:20230612T143000
DTEND;TZID=GMT:20230612T144500
DESCRIPTION:Since the deployment of the first satellite equipped with a Syn
 thetic Aperture Radar (SAR) into orbit in 1978\, the use of SAR imagery ha
 s been a vital part of several scientific domains\, including environmenta
 l monitoring\, early warning systems\, and public safety.\nSAR could be de
 scribed as "non-literal imaging" since the raw data does not resemble an o
 ptical image and is incomprehensible to humans.\nFor this reason\, raw dat
 a is typically processed to create a Single Look Complex (SLC) image\, whi
 ch is a high-resolution image of the scene being observed. The processing 
 of raw data to create a SLC image involves several steps\, including range
  compression\, Doppler centroid estimation and azimuth compression.\nProce
 ssing raw data requires a significant amount of computer power\; as a resu
 lt\, it is almost never practical to do it on board. As a direct consequen
 ce\, the data is transmitted back to Earth to be processed.\nThe objective
  of next-generation studies [1] is to optimize Earth Observation (EO) data
  processing and image creation in order to deliver EO products to the end 
 user with very low latency using a combination of advancements in the on-b
 oard parts of the data chain.\nIn this work\, we focus on a sea scenario a
 nd propose to eliminate any pre-processing by training a Deep Convolutiona
 l Neural Network (DCNN) to directly recognize bright targets on raw data. 
 \nThis indeed might substantially shorten the delivery time thus improving
  the efficiency of satellite-based maritime monitoring services.\nIn this 
 regard\, the availability of training data represents one of the critical 
 issues for the development of machine learning algorithms. In fact\, the e
 fficacy of the final machine learning-powered solution for a specific appl
 ication is ultimately determined by the quality and amount of the training
  data.\nHowever\, to date\, there are no training SAR raw data available i
 n scientific literature with regard to the specific topic of sea scenario 
 monitoring. Furthermore\, their generation from real data is a time-consum
 ing task.\nIn this work we propose and investigate physically and statisti
 cally based approaches to simulate a marine scenario and generate realisti
 c synthetic training SAR raw datasets.\nWe then trained and evaluated a st
 ate-of-the-art DCNN on the generated synthetic dataset and successively on
  real raw data extracted from ERS imagery archive. It is one of the first\
 nexperiments proposed in the SAR literature and results are quite encourag
 ing\, as they reveal that a well-trained DCNN can correctly recognize stro
 ng scattering objects on SAR raw data. \n\n[1] M. Kerr\, et al. “EO-ALER
 T: a novel architecture for the next generation of earth observation satel
 lites supporting rapid civil alerts”\, in 71st International Astronautic
 al Congress (IAC)\, 2020.\n\nAcknowledgments \nThis work was carried out i
 n the framework of the APP4AD project (“Advanced Payload data Processing
  for Autonomy & Decision”\, Bando ASI “Tecnologie Abilitanti Trasversa
 li”\, Codice Unico di Progetto F95F21000020005)\, funded by the Italian 
 Space Agency (ASI). ERS data are provided by the European Space Agency (ES
 A).
DTSTAMP:20260427T145242Z
LOCATION:Sala Biblioteca @ PoliBa
SUMMARY:Bright Target Detection on SAR Raw Data Based on Deep Convolutional
  Neural Networks - Giorgio Cascelli\, Alberto Morea\, Khalid Tijani\, Nico
 lò Ricciardi\, CATALDO GUARAGNELLA\, Raffaele Nutricato
URL:https://talks.staging.osgeo.org/foss4g-it-2023/talk/URR3NB/
END:VEVENT
END:VCALENDAR