DESIGN AND EXPERIMENT OF A CORN INTER-PLANT WEEDING MACHINE BASED ON VISUAL RECOGNITION

To address the challenges associated with high interrow weeding difficulty and seedling damage in corn fields, a weed removal machine based on crop spacing recognition was designed. It captures crop images at variable speed intervals, obtains corn seedling centroid coordinates via image stitching and skeleton extraction, calculates actual plant spacing through pixel-to-real coordinate transformation, and enables real-time control of the weeding device. Key components were analyzed for motion trajectories and critical parameters. Field tests revealed optimal performance at 0.45 m/s, namely, a 92.6% weed removal rate with 2.05% seedling damage, meeting operational requirements. This research provides technical and equipment support for interrow weeding.