• 20141113-M0274_13

RGB Imaging Sensor

RGB imaging remains the workhorse of digital phenotyping sensors producing rapid, easy to interpret data and vital phenomic parameters on morphology, biomass, plant growth, and architecture and (de-) coloration.

WPS has extensive experience with both the soft- and hardware that’s needed to get the best results from a RGB plant imaging setup.

Some facts and figures:

  • WPS has 150 RGB imaging sensors operational all over the world
  • Daily > 1,000,000 plants are being analyzed with WPS sensors and algorithms
  • WPS RGB sensors and cabinets can be used for different plant sizes, from seedlings to mature plants of all major agronomical crops.

It all starts with image quality
Relevant information is only acquired via good and objective image quality. WPS uses advanced optics and sensors, a finely balanced setup that leads to the best possible image with reliable objective color information. This is essential for the next step, image analysis.

Image analysis
WPS has extensive experience with image analysis of plants and has developed a wide range of algorithms for this purpose, to quantify morphologic traits such as

  • Digital biomass
  • Height
  • Convex hull
  • Centre of mass/ gravity
  • Projected surface
  • Filling
  • Stem diameter
  • Number of leaves
  • Flowering surface
  • Senescence
  • …and many others.

Color regions and histogram
WPS has developed a user friendly intuitive tool for selecting color (regions) of interest. This enables the user to correlate color information to certain experimental effects in plants and apply it thereafter in an automated fashion. It can, for example, quantify the greenness of the plant, in fertilizer optimization, or quantify progress of senescence.

Alongside specific color regions identification and quantification, histogram values can be calculated and utilized to elucidate early correlations between traits/ effects and colour changes within the plants.

Advanced machine learning
WPS also has several options to use state-of-the-art machine learning algorithms to detect features that are difficult to extract when using conventional image analysis techniques. In the movie below, we demonstrate an ability to automatically count bud and/or flowers and are currently being extended to spikes and other anatomical features.

Watch video of Feature Tracking 3D