Novel Hyperspectral Sensors for Quantitative Assessment of Health and Nutrition States of Crop Plants

Quantitative assessment of phenotypic properties of crop plants in relation to different genotypes, nutrition, stress tolerance, and fruit quality has become increasingly important in crop plant research, modern plant breeding, and particularly in precision agriculture / smart farming. Hyperspectral imaging linked to subsequent computational intelligence based analysis has proven its suitability to unravel complex information in a number of different application areas, such as geology, defence, etc. The extension of this approach to crop plant research, plant breeding, agriculture, and food processing has started just quite recently.
Hyperspectral imaging has paved the way to obtain image data across wide wavelength ranges in selective spectral bins. Using state-of-the-art sensors typically leads up to several hundred spectral images of the same object or scenery. Every acquired image contains different information depending on the reflection/absorption properties of all objects within the field of view of the camera. In the case of plant or tree leaves these properties are mainly determined by biochemical compounds of tissues near to the leaf surface and by composition and morphology of the leaf surface itself.

This talk will briefly introduce the physical basis and demonstrate characteristic properties of a recently developed screening system along with extensive data from systematic field trials obtained between 2010 and 2015.

Udo Seiffert is a Research Associate at the University of Magdeburg where he works on machine learning for image processing and pattern recognition, growing multi-dimensional Self-Organizing Maps (SOM), applications in motion tracking, trac surveillance, and Advanced Driver Assistance Systems (ADAS).

Udo Seiffert is Head of Department "Biosystems Engineering" at the "Fraunhofer Institute of Factory Operation and Automation" Magdeburg, Germany and Professor for "Artificial Intelligence" at the University of Magdeburg. His research interests are primarily image and signal processing with statistical and machine learning based methods for applications in agriculture, crop plant breeding, and food processing.