Abstract: The acquisition of information regarding fruit tree canopy structure is essential for optimizing variable spray operations. In order to enable realtime data collection on canopy structure, a twodimensional (2D) ground laser radar was employed to scan fruit trees with varying Leaf Area Index （LAI） in this study. Threedimensional (3D) point cloud images of the fruit trees was generated by using MATLAB software, and was consistent with the actual tree morphology, which indicated a strong correlation between the point cloud data and structural information of the fruit tree. The results revealed that the relative errors of tree height and tree width measured by utilizing 2D LiDAR were 2.22% and 4.11%, respectively. However, the accuracy of tree thickness was found to be influenced by LAI, the relative measurement errors for thickness increased from 5.3% to 41.1% when LAI increased from 0 to 3.68. The predictive models for LAI were developed for frontal and dorsal scans, with equations of y=1.265x-0.313 7 and y=1.230 5x-0.338, respectively. Ftest results highlighted significant differences among  the samples, with model goodness of fit exceeding 0.9. This study offers valuable technical insights and model support for decisionmaking in orchard variable spray operations.

Key words: fruit tree, variable spray, leaf area index, LiDAR, point cloud

摘要： 果树冠层结构信息获取是变量喷雾作业重要前提，为实现果树冠层结构信息在线采集，采用二维地面激光雷达对不同叶面积指数的果树进行，并运用MATLAB软件绘制果树的三维点云图。绘制出的果树三维点云图与实际果树形态的一致，表明点云数据与果树结构信息具有高度相关性。研究结果表明采用2D LiDAR测量树高与树宽的相对误差分别为2.22%和4.11%，但树厚的精度与LAI相关，LAI由0增加到3.68时其测量相对误差由5.3%增加到41.1%。拟合出正面和背面扫描时的LAI预测模型分别为y=1.265x-0.313 7和y=1.230 5x-0.338， F检验结果显示样本间存在显著性差异，且其模型的拟合优度均大于0.9。该研究可以为果园变量喷雾决策提供技术与模型支撑。

关键词: 果树, 变量喷雾, 叶面积指数, 激光雷达, 点云