基于地基激光雷达点云的植被表型特征测量

Plant phenotypes are the result of the combined effects of genotypes and the external environment. Accurate measurement of plant phenotypes is of great significance for the study of plant physiological characteristics and functional traits. In this study, Solidago canadensis was taken as the object, and 20 plants were cultured indoors for 3 months. Each month, the experimental plants were scanned at multiple stations and the point clouds were fused using a terrestrial LiDAR scanning (TLS) system to achieve continuous observation of the plant growth process. For the discrete point clouds obtained by scanning, the plant height was obtained using the multi-endpoint three-dimensional coordinate reconstruction method, and the leaf surface was reconstructed based on the Delaunay triangulation network of the leaf point clouds to obtain structural parameters such as the true height, leaf area, leaf inclination angle and azimuth angle of the plant. Comparing with the manual measurement results, it was found that for the plant height obtained by point cloud reconstruction compared with the real plant height, the coefficient of determination (R²) of the similarity between the two was 0.991, and the R² of the similarity of leaf area, leaf inclination angle and azimuth angle were 0.989, 0.949 and 0.871 respectively; the non-destructive plant phenotype measurement was achieved based on the TLS point cloud reconstruction method, and high-precision plant phenotype characteristics could be obtained; multi-temporal scanning can accurately monitor the changes in phenotypic characteristics during the plant growth process.

植物表型是基因型与外界环境共同作用的结果。精确测量植物表型对于植物生理特征与功能性状研究具有重要意义。本研究以加拿大一枝黄花(Solidago canadensis)为对象,对20株植株进行3个月室内培养,各月利用地基激光雷达扫描(terrestrial LiDAR scanning,TLS)系统对实验植株进行多站扫描和点云融合,实现对植株生长过程的连续观测。对于扫描获取的离散点云,利用多端点三维坐标重构法获取植株高度,并基于叶片点云的Delaunay三角网重构叶片表面,获得植株的真实高度、叶面积、叶倾角和方位角等结构参量。对比手动测量结果,发现基于点云重构获得的植株高度与真实植株高度对比,二者间相似性的决定系数(R~2)为0.991,叶面积、叶倾角、方位角相似性R~2分别为0.989、0.949和0.871;基于TLS点云重构法实现了非破坏性的植物表型测量,能够获得高精度的植物表型特征;多时相扫描能精确监测植物生长过程的表型特征变化。