关键词:
无人机
激光雷达
湿地松
生长性状
遗传分析
摘要:
近年来,随着林业信息化和智能化的推进,无人机在林业数据采集和分析中的应用日益广泛。本研究以湿地松为研究对象,对比无人机RGB相机(可见光相机)和LiDAR相机(激光雷达相机)在获取湿地松生长性状与地面真实值之间的精度差异,并借此方法寻找湿地松优良家系。这两种无人机通过执行相同的飞行路线,飞行后数据经过数据分析得到每株湿地松的树高和冠幅,并建立胸径预测模型,比较两个相机的预测精度和选育优良家系。结果显示,LiDAR相机在50 m和80 m高度下,树高预测的相关系数R2均为0.88,优于RGB相机的预测精度。50 m高度下RGB相机倾斜摄影的相关系数R2为0.86,优于正射影像的0.78。基于树高和冠幅的胸径预测模型也表现出较高的精度。不同飞行任务获取的数据均对选育优良家系有积极的作用,与真实选育结果一致。虽然RGB相机的预测精度略低于LiDAR相机,但差异不显著,显示其可作为低成本替代方案。激光雷达的飞行高度(50 m和80 m)对树高测量精度无显著影响,80m具有较短的飞行时间,RGB相机倾斜摄影的精度高但飞行时间较长。湿地松的树高、胸径和冠幅均受中高遗传力控制,育种值分析表明10号家系表现优异,适合进一步选育。本研究为无人机在林木资源调查和表型遗传分析应用中提供了科学支持。In recent years, with the advancement of forestry informatization and automation, the application of UAV in forestry data collection and analysis has grown increasingly prevalent. This study focuses on Pinus elliottii to compare the accuracy of UAV RGB sensor (Visible Light Camera) and LiDAR sensor (Lidar Camera) in capturing growth traits relative to ground truth measurements, aiming to identify superior Pinus elliottii families using these methods. Both UAV followed the same flight paths, and after data processing, tree height and crown width for each tree were obtained. A diameter at breast height (DBH) prediction model was then developed, and the accuracy of the two sensors’ predictions was compared, along with their utility in family selection. Results showed that the LiDAR sensor achieved a correlation coefficient (R2) of 0.88 for tree height predictions at both 50 m and 80 m altitudes, outperforming the RGB sensor. At 50 m, the RGB sensor’s oblique photography produced an R2 of 0.86, superior to the 0.78 obtained from orthophotos. The DBH prediction model, based on tree height and crown width, also exhibited high accuracy. Data collected from different flight missions positively contributed to the selection of superior families, aligning well with actual breeding outcomes. Although the RGB sensor’s prediction accuracy was slightly lower than that of the LiDAR sensor, the difference was not significant, indicating that it ca