Detection of perennial weed patches in ripening cereals with an image analysis approach
- Publikations-Art
- Kongressbeitrag
- Autoren
- Tobias Dittmann and Martin Weis and Christoph Zecha
- Erscheinungsjahr
- 2012
- Veröffentlicht in
- Proceedings of the 12th International Conference of Agricultural Engineerig CIGR 2012
- Seite (von - bis)
- C2186
- Tagungsname
- CIGR 2012
- Tagungsort
- Valencia, Spain
- Tagungsdatum
- June 10--15
On agricultural fields weeds are not homogeneously distributed but show patchy patterns. Particularly perennial weeds occur in patches that are stable over several years. Due to this characteristic perennial weeds can be controlled site-specifically. Site-specific chemical weed control offers substantial herbicide savings. Furthermore, mechanical control measures can be conducted to weed patches, saving time and money. However, site-specific weed control requires reliable automated recognition and mapping of weed patches. Little research has been done on the recognition of weeds at lactitc ripeness, therefore images were taken during the harvest period. An approach is presented to recognize and classify the perennial weeds Rumex obtusifolius, Cirsium arvense and Convolvulus arvensis based on color and shape features of the weeds flower. It can be concluded, that these weed species can be recognised within a cereal field at lactic ripeness and distinguished from each other with image processing
and the developed automated recognition algorithms. The approach can be used to map perennial weed patches within cereal fields, providing information for site-specific weed control.