Multi-wavelength laser line profile sensing for agricultural crop characterization

Wolfram Strothmann; Arno Ruckelshausen; Joachim Hertzberg

In: Proceedings of SPIE Vol. 9141: Optical Sensing and Detection III. SPIE Photonics Europe (SPIE-2014), April 14-17, Vol. Vol. 9141, 4/2014.


Triangulation-based laser line profile sensing is a widely used technique for precise and fast 3D measurement in industrial applications. More recently, it is used for applications in the area of outdoor agricultural sensing as well. However, in many agricultural applications high resolution range data by itself is insufficient. There is also a strong need for local spectral intensity information. As a consequence the combination of image-based range scanners with spectral imaging is often necessary. Under varying environmental conditions this is a tedious and error-prone problem, though. In contrast, the novel approach shown here allows capturing range data along with spectral laser reflectance and pixel-wise backscattering information at multiple, selectable wavelengths using a single sensor system. The system consists of multiple continuous wave (CW) line lasers simultaneously captured by a single monochrome imager. A system ready to capture 3 line lasers at 100 Hz was set up. Line lasers at different wavelengths in the visible and NIR range can be combined in accordance with the requirements of a specific application. Consecutively captured images are matched using sum of absolute differences (SAD) in order for tracking relative movement between the sensor system and the analyzed object. This allows normalizing images before the evaluation of reflectance and scattering. Furthermore, the SAD-based matching is used for accurate assembly of range and reflectance information gathered from different laser lines. It results in 3D point clouds with spectral laser reflectance and backscattering information at multiple, selectable wavelengths available for each point.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence