Abstract: Visualization of simulated crop growth and development is of significant interest to crop research and production.  This study aims to address the phenomenon of organs cross-drawing by developing a method of collision detection for improving vivid 3D visualizations of virtual wheat crops.  First, the triangular data of leaves are generated with the tessellation of non-uniform rational B-splines surfaces.  Second, the bounding volumes (BVs) and bounding volume hierarchies (BVHs) of leaves are constructed based on the leaf morphological characteristics and the collision detection of two leaves are performed using the Separating Axis Theorem.  Third, the detecting effect of the above method is compared with the methods of traditional BVHs, Axis-Aligned Bounding Box (AABB) tree, and Oriented Bounding Box (OBB) tree.  Finally, the BVs of other organs (ear, stem, and leaf sheath) in virtual wheat plant are constructed based on their geometric morphology, and the collision detections are conducted at the organ, individual and population scales.  The results indicate that the collision detection method developed in this study can accurately detect collisions between organs, especially at the plant canopy level with high collision frequency.  This collision detection-based virtual crop visualization method could reduce the phenomenon of organs cross-drawing effectively and enhance the reality of visualizations.

Key words: wheat ,  collision detection ,  bounding volume hierarchy ,  virtual plant ,  morphology