Biomass estimation from images for greenhouse experiments
- Publication Type
- Journal contribution
- Authors
- Martina Keller and Martin Weis and Roland Gerhards
- Year of publication
- 2010
- Published in
- Proceedings 16. Workshop Computer-Bildanalyse in der Landwirtschaft
- Editor
- Manuela Zude and Martin Kraft
- Pubisher
- Leibnitz Institute for Agricultural Enineering (ATB) , Potsdam-Bornim/Braunschweig
- Band/Volume
- 73/
- Series/labeling
- Bornimer Agrartechnische Berichte
- ISBN / ISSN / eISSN
- 0947-7314
- Page (from - to)
- 13-20
- Conference name
- 16. Workshop Computer-Bildanalyse in der Landwirtschaft
- Conference location
- Braunschweig
- Conference date
- 4.05.2010
Chemical weed control is a common practise in agriculture. A reduction of the amount of herbicides can be achieved using site-specific applications. To develop models for an optimized herbicide use on a sub-field level, dose-response relations between herbicides and weed species need to be derived. Whole-plant bioassays were conducted in the greenhouse with several weed species to determine the response to different herbicides. The tested herbicides were applied at different rates and the response was measured. Above-ground dry matter is usually measured at the end of such experiments to quantify the overall effect of the herbicides on plant growth. In our experiments, additionally red and infrared images of the plants were taken before harvest. From difference images, generated by subtracting red from infrared channel, the coverage of the green/photosynthetically active plant parts could be calculated using image processing techniques. Coverage was correlated with the dry matter values, resulting in significant correlations. Dose-response curves were fitted to the measured dry matter and coverage. The conducted experiments confirm that this technique can be used to sample data for bioassays and dose-response experiments. Such automation eases bioassay experiments compared to classical dry matter measurements.