09 Multispectral Time-of-Flight Range Imaging: Towards a 3.5 D Camera

Solèr Ursin, Merbold Hannes, Iten Marc-Alexander, Catregn Gion-Pol

  1. University of Applied Sciences of the Grisons, Institute of Photonics and ICT (IPI), Chur, Switzerland

In recent years, TOF imaging has been used for a variety of applications such as object detection or gesture recognition. Commercially available TOF cameras typically employ near-IR LEDs, such that the illumination is invisible to the human eye. This has the disadvantage that no color, or in other words no spectral information, can be resolved. Dual camera approaches using a TOF and a regular RGB camera provide nice spectral data but result in high costs and a complex data processing scheme. We present a multispectral TOF imaging approach that circumvents these limitations by using a sequenced multi-wavelength illumination scheme from multiple groups of LEDs with different center wavelengths. An alternative approach using swing-in narrow bandpass filters within the bandwidth of the default near-IR LEDs is also discussed. We refer to our systems as 3.5 D cameras as both approaches allow recording three spatial dimensions as well as a limited number of spectral data points. Evidently, combining the data into multispectral TOF images (i.e. 3.5 D datasets) allows recording the object distance as well as discriminating spectral properties. Since the approaches can easily be extended to other spectral ranges, they open up a wealth of interesting applications, e.g., the simultaneous resolution of surface shape and chemical composition of arbitrary samples.