Multispectral data interpretation

Multispectral analysis provides information about object properties

Multispectral laser scanning: Moisture measurement on a tunnel wall
© Fraunhofer IPM
Moisture on a tunnel wall captured via multispectral laser scanning: The grey values in the high density 3D point represent backscatter intensity.
Multispectral laser scanning: Moisture measurement on a tunnel wall
© Fraunhofer IPM
The same data set with colored areas encoding surface moisture (red: high, blue: low)

When measuring objects in 3D, multispectral analysis often makes it possible to assign further parameters to the object in addition to the pure geometric data. This is done by taking advantage of the fact that each material has a unique spectral fingerprint. This means that light of different colors is absorbed to different degrees by objects with varying properties. For multispectral analysis we use either passive systems, i.e. cameras with corresponding color filters, or active systems, i.e. laser scanners with laser beams of different wavelengths. Analyzing a multispectral signal allows, for example, to draw conclusions about the health condition of plants, to distinguish between ice and water and to make moisture in rooms visible.

Reliable moisture detection

Water is a good example: while it is transparent in the visible spectral range, it exhibits strong absorption in the near infrared (NIR) at optical wavelengths of 1450 nm and 1950 nm. This property can be used to detect water. For this purpose we use Differential Optical Absorption Spectroscopy (DOAS). DOAS is common practice, for example, for determining moisture in food. Two collinear laser beams of different wavelengths scan the object surface. The beams are very specifically absorbed by water. At a wavelength of 1450 nm, the measuring beam is selected according to the absorption band of water. The second laser with 1300 nm wavelength is located outside this absorption band and serves as an intensity reference. The moisture value results from the intensity analysis of the two signals. The recorded intensity values of both wavelengths are standardized to the target intensities of an ideal dry target with a Lambertian scattering behavior and a reflectivity of about 90 percent. This makes it possible to differentiate very precisely between different moisture values (e.g. strong, medium, weak) while taking into account the reduction in overall intensity caused by material changes, color, dirt, etc.

Green gap: Absorption spectrum of chlorophyll
© Fraunhofer IPM
The spectral range between 490 and 620 nm in the absorption spectrum of chlorophyll is called "green gap". This is why plants appear green to us.
Normalized Difference Vegetation Index (NDVI) for assessing the health of vegetation
© Fraunhofer IPM
The Normalized Difference Vegetation Index (NDVI) is the most common index for determining the general health of vegetation. It compares the low reflection of plants in the red spectral range with the high reflection in the near infrared (NIR) range.

Green detection and condition monitoring of plants

Plants also possess a special spectral fingerprint. They usually show the so-called green gap. Due to a high chlorophyll concentration, plants absorb very weakly in the green spectral range, while absorption is strong in the blue and red spectral range. This is why plants appear green to us. They also show a special feature in the reflection of light: red light is reflected particularly weakly, while light in the near infrared (NIR) is reflected strongly. The healthier the plant, the stronger the reflection in the NIR. By forming the ratio of the measured reflection, the state of health of a plant can therefore be readily determined. Depending on the plant species, however, reflection is slightly shifted in the wavelength range or the characteristic reflection in the blue or green spectrum is more pronounced. For this reason, a large number of vegetation indices have become established, which are appropriate depending on the plant species.

Applications »Multispectral data analysis«

 

Efficient and sustainable vegetation maintenance

A multispectral camera-based system for automated green detection enables the sensor-controlled application of herbicides for vegetation control in rail networks.

 

Detecting moisture and vegetation in tunnels

Our optical systems for tunnel inspection are based on multispectral phase scanning. They provide data on the 3D geometry, surface structure, and moisture of tunnel constructions.