Multispektral dynamisk 3D-avbildning med laser - Slutrapport

Authors:

  • Håkan Larsson
  • Elias Amselem
  • Jimmy Berggren
  • Carl Brännlund
  • Philip Engström
  • Frank Gustafsson
  • Fredrik Hemström
  • Markus Henriksson
  • Christina Grönwall
  • Anna Linderhed
  • Ove Steinvall
  • Gustav Tolt
  • Lars Sjökvist

Publish date: 2013-12-31

Report number: FOI-R--3768--SE

Pages: 43

Written in: Swedish

Keywords:

  • High resolution 3D imaging
  • multi-spectral
  • dynamic
  • laser sensors
  • signal
  • processing
  • modeling

Abstract

Laser sensors can generate data with three dimensions and a spectral description of objects and surroundings. This can be very useful in military operations for target recognition and as basis data for situation awareness. Scanning laser systems are of great use for mapping, in preparation phase or during missions. Increase in data acquisition speed is needed for some tasks and therefore often array detectors are used to measure multiple points with each laser pulse. Particularly interesting is the system that will build and use a localization model during movement, so-called SLAM (Simultaneous Localisation and Mapping). SLAM can be performed with good quality using simple sensors, which pulls down the cost of otherwise expensive sensor systems. As the sensors are shrinking and they soon can be placed on smaller autonomous platforms, such as UAV -type multi-rotors, which will make autonomic mapping cheaper and available for use in areas of conflict. By repeated mapping at different occasions, change detection algorithms can be used that effectively reveals potentially threatening activities. Survey of moving targets requires that the entire scene is measured the same time. This can be done with 3D imaging array detectors, which also allows the collection of video data. Prototype systems for this, based on different techniques, are developed at several places in the world. Performance regarding the number of pixels in the sensor and sensor robustness evolve towards operationally useful systems. Important for recognition of targets behind camouflage is that the range resolution in each pixel is such that several surfaces that partially conceal each other can be distinguished, also when they are separated by a few centimeters. Photon counting laser radar shows promising performance for this and array detectors that allow imaging in video rate begins to be developed, as yet however, with relatively few pixels. The possibility of recognition of material is possible with multispectral measurements where the reflectance of surfaces is measured at several wavelengths. This can be done either by active imaging using lasers or by or by passive imaging using the background light from the sun. Differences in spectral characteristics can e.g. be used to distinguish people and vehicles from the surroundings (vegetation or urban materials), identify objects that people hold in their hands or re- identification of people in a crowd depending on the clothes. Especially interesting are the reflectance properties outside the visible wavelength range, as often less work with signature adaptation has been performed in those regions. The technology of multispectral laser radar is still far from operational use and new concepts for data collection need to be developed to reduce analysis time. As 3D imaging multispectral laser radar yet cannot be performed on an operational reasonably way, it is important to study the sensor fusion of 3D data and 2D imaging multispectral data.

Share page on social media