Imaging laser sensors- Final report for the period 2017-2019

Authors:

  • Per Jonsson
  • Maria Axelsson
  • Lars Allard
  • Linnéa Axelsson
  • Fredrik Bissmarck
  • Christina Grönwall
  • Julia Hedborg
  • Markus Henriksson
  • Max Holmberg
  • Fredrik Kullander
  • Pontus Köhler
  • Patrik Lif
  • Jonas Nordlöf
  • Hannes Ovren
  • Mattias Rahm
  • Lars Sjökvist
  • Daniel Svedbrand
  • Ulf Söderman
  • Gustav Tolt
  • Michael Tulldahl

Publish date: 2019-12-17

Report number: FOI-R--4874--SE

Pages: 36

Written in: Swedish

Keywords:

  • 3D imaging
  • laser radar
  • photon counting
  • panoramic imaging
  • UAV
  • underwater imaging
  • target detection
  • change detection
  • machine learning

Abstract

This report summarizes and describes the results of the three-year research project Imaging laser sensors during the period 2017-2019. This work has been funded by the Swedish Armed Forces R&D programme for Sensors and low observables (FoT SoS, AT.9220419). The project has studied photon-counting sensor systems for long ranges and for underwater imaging, as well as a 3D imaging laser sensor systems mounted on a UAV (unmanned aerial vehicle). In addition to results, the report also describes the collaboration FOI has had with others and the dissemination of knowledge to the Swedish Armed Forces, FMV and other stakeholders. Imaging laser systems with photon-counting technology can be made relatively small and compact since lasers with relatively low pulse energies can be used. In addition, you can measure scenes with large depth with high range resolution. We have developed a photon counting sensor system and tested it at ranges over 2 km. The system is based on an array detector that has 128 × 32 pixels where each pixel measures the time-of-flight of the laser pulses. Although the detector is large for photon-counting sensors, it is small compared to image-generating sensors in visual and IR. Therefore, we have developed a panoramic technique so that the system can cover larger scenes and tested it in field trials both in sunlight and in darkness. We have developed methods for reducing noise, detecting surfaces of objects in data, combining image data from multiple views and detecting targets with change detection. We have also performed initial experiments with photon-counting systems for underwater imaging. A 3D imaging laser sensor system on a UAV has the ability to look down through obscuration. In addition, there is more information about a target in 3D than in 2D. In the project we have been developed our aerial system so that more data can be processed on board the UAV. We have shown that the sensor system can detect vehicles under trees, which are difficult for passive sensors to detect. Data collected with the system has been used to test methods for scene analysis based on machine learning with semantic segmentation, a classification of each point in the data.