Morphological image analysis applied to crop field mapping

Previous research about automatic vision systems applied to crop field imaging fails when the scene is too complex, in the sense that the individual plants are connected together or to weed patches. In this paper, we show that morphological image processing can efficiently solve these problems provided that a priori knowledge about the shape of the plants is available. For example, vein networks appearing within the target plant leaves allow us to automatically mark and segment the plants. Results are shown on a series of outdoor scenes of a test field. (C) 2000 Elsevier Science B.V. All rights reserved.