Erkennung von Beikraut in Sonderkulturen durch Klassifikation von Multisensorinformation

Conference paper

Links:
Authors:

Johannes Meyer
Christian Frese
Christian Frey

Source:

Manuela Zude-Sasse, Martin Kraft (eds.), Workshop Computerbildanalyse und Sensorik in der Landwirtschaft, Bornimer Agrartechnische Berichte, vol. 88, Leibniz-Institut für Agrartechnik Potsdam-Bornim e.V., 2014.

Pages:

191-199

Conference:

20. Workshop Computerbildanalyse und Sensorik in der Landwirtschaft, Osnabrück, May 26, 2014

This contribution addresses the discrimination of plants and weed in specialized cultivations. The proposed method enables automatic weed control by an autonomous robot. A multi-sensor system captures image and range data of the plants. Features based on color, texture, and range information are computed. A classifier is trained to distinguish the cultivated plants from weed. The classification performance is evaluated on real data acquired in a boxwood cultivation.