Image matching algorithm for weed control applications in organic farming

In general a weed control robot system comprises a weed classification and a weed destruction unit. Both unit are physically separated from each other inside the robot. For that reason the weed classification unit cannot inspect the weed destruction result. Hence an important step for the destruction of weed in organic farming with a robot system is the tracking of classified weed positions. In this paper the tracking is realized with a low-resolution VGA camera and a developed matching algorithm. The tracking is necessary as the robot position can move due to stones or other obstacles on its path. Also speed changes of the robot lead to weed destruction problems. The performance of the proposed image matching algorithm is illustrated by simulation and experimental results.