【摘要】 为了快速、无损、自动获取玉米植株的三维信息,设计了一种基于移动机器人平台的玉米植株三维信息采集系统。首先,通过四轮驱动移动机器人与升降平台配合,精准控制Xtion深度相机多视角采集盆栽玉米植株的三维点云数据;然后,对获取的点云进行配准拼接与滤波,实现玉米植株的三维重建;最后,对重建后的玉米模型进行叶片的分割与参数测量。该系统的移动机器人运动误差可以控制在1 cm之内,玉米植株叶片参数测量误差在1%～5%之间,证明了该系统进行玉米植株三维信息采集的可行性。更多还原

【Abstract】 Plant 3D information is an important parameter in the process of plant growth,reflecting the normal growth state of plants. In order to quickly and non-destructively acquire the 3D information of corn plants,a 3D information acquisition system based on mobile robot platform was designed. The fourwheel-drive robot was equipped with a lifting platform and Xtion camera at the end,and the Xtion camera acquired a 3D point cloud at multiple angles through the motion control of the robot. Firstly,the hardware structure and software platform of the acquisition system were described. Then,the steering of the four-wheel drive robot simplified the kinematics analysis. According to the system collection needs,the four-wheeled robot and the lifting platform controlled the camera to make a circular motion with a radius of R. The camera collected a 3D point cloud at interval of θ until the robot was moved for one circular motion. Finally,according to the transformation matrix T for calculating the point cloud of the adjacent angle,the registration and splicing of the 3D point cloud were performed,and the filter was used for filtering to complete the 3D reconstruction of the corn. The watershed algorithm was used to mesh and segment the reconstructed maize plants,and the leaf length parameters were measured. The results showed that the motion error for mobile robots was less than 1 cm. The error between predicted value and the true value of the leaf length of corn plant was between 1% and 5%. The system provided a new way for the collection of 3D information of corn plants.更多还原