基于旋翼无人机平台的作物冠层表型特征与光截获功能的定量化与评估

The development of a high-throughput phenotyping platform shows great significance to the new varieties breeding and cultivation management optimization. It is high-efficiency, less disturbance and low cost to monitor the growth and development of crops with the high-throughput phenotyping platform in the field. Using unmanned aerial vehicle (UAV) carrying different sensors (e.g. RGB and multi-spectral cameras) to capture image sequence, we plan to develop a high-throughput method to monitor the dynamic changes of maize and soybean monoculture and intercropping canopy in the field. The ortho-mosaic and the three dimensional (3D) point cloud of the whole scene can be reconstructed through the development and application of image interpretation models, then phenotyping parameters will be extracted based on the ortho-mosaic and the reconstructed 3D structural model of crop canopies based on 3D point cloud. The light distribution model coupled with the canopy structural models is used to estimate the light distribution of different planting patterns in the field. Compared the extraction results with the parameters extracted from the structural models based on the 3D digitalization and the corresponding values based on the measurement of the phenotyping parameters in the field to evaluate the accuracy of the platform, and to do the systematic evaluation of the functional-structural plant model which reconstructed based on the high-throughput platform by comparative analysis of the canopy light distribution values between the estimations (from the light distribution model based on the high-throughput platform and the precise structural model, respectively) and the measurement. The study can provide important theoretical and practical application values for breeding and precision agriculture practices.

开发高通量植物表型测定平台，对作物新品种选育和栽培措施优化具有重要意义。基于无人机平台监测大田作物生长具有高通量、干扰少、低成本等优势。拟采用基于低空无人机的监测平台，搭载RGB相机和多光谱相机对玉米和大豆单作和间作种植模式不同生育期的冠层进行监测，获取多源图像数据。通过应用和开发图像解译模型，基于重建的二维拼接图像和三维点云提取冠层表型参数并重建冠层三维结构模型。将冠层结构模型与已开发的冠层光分布模型耦合，实现不同种植模式的大田冠层光分布的定量化。将基于无人机平台提取的冠层表型参数值与基于三维数字化数据构建模型获得的参数值、大田实测值比较，评估高通量表型测定平台的精度；将所构建的冠层光分布模型的模拟结果与基于精确冠层结构数据的冠层光分布模拟结果、大田光分布实测结果比较，实现对该平台获取的冠层结构-功能的系统地评估。本研究对现代育种与精准农业实践等相关领域均具有较重要的理论和应用价值。

现代作物新品种选育和农田作物精确管理都亟需有快速、精确的植物表型参数获取方法支持。基于无人机平台监测大田作物生长状况具有高通量、低成本等优势。本研究进行了不同品种、不同行配比的玉米/大豆间作和单作的大田实验。采用轻量型无人机获取不同生育阶段的单作、间作群体冠层的多视角图像。为克服无人机搭载的相机精度不够而导致重建冠层点云精度不高的问题，采用了更新一代的无人机，基于所搭载的更高精度相机并采用超低飞行高度，成功获取了多视角高精度图像序列。基于图像序列重建冠层点云后，采用了噪点剔除和畸形面片消除算法，比较理想地实现了作物冠层三维结构模型的构建。由此实现了所构建的冠层三维结构模型与冠层光分布模型的耦合，并基于不同生育阶段间作群体冠层光分布的实测值验证了冠层光分布模拟模型。将冠层光分布模型的模拟结果与基于精确冠层结构模型的冠层光分布模拟结果、大田光分布实测结果比较，实现了对基于无人机获取的冠层结构模型的系统性评估。同时，还建立了高效获取作物株高、植株个体叶面积、群体叶面积剖面分布及器官尺度表型参数的方法，将基于无人机平台提取的冠层表型参数值与基于三维数字化数据构建模型获得的参数值、大田实测值比较，评估了所建立方法的精度，评估结果较理想。此外，还采用无人机搭载多光谱相机对大田冠层进行航拍，基于多源图像数据实现了冠层结构建模。本研究对现代育种与精准农业实践等相关领域均具有较重要的理论和应用价值。