Abstract:Leaf area index (LAI) is an important crop phenotyping parameter and an important indicator of crop growth and yield. Using vehicle-mounted three-dimensional (3D) LiDAR, crop morphological parameters such as plant height and LAI can be quickly obtained. Maize was taken as the research object, and a method of calculating LAI based on the ratio of the number of stratified point clouds or the number of stratified point clouds to the number of ground point clouds was proposed by using the data of three-dimensional LiDAR point clouds in vehicle. 3D point cloud data of Jingnongke 728 and Nongda 84 were obtained by vehicle platform. Firstly, the point cloud data were preprocessed to obtain the point cloud data of the measured LAI true value region. Secondly, the point cloud of maize plant and the ground point cloud were segmented. According to the fluctuation degree of the ground, the distance threshold of random sample consensus’s plane model was set to be 0.06m. then according to the vertical structure distribution of maize, the maize plants were divided into high, middle and lower layers, and each layer was calculated. The number of clouds was marked as H, M and L, respectively. At the same time, the ratio of the number of point clouds in each layer of high, middle, and lower layers to the number of ground point clouds was marked as Hr, Mr and Lr. Finally, the linear regression models of the true values of H, M, L and Hr, Mr, Lr and LAI were established respectively. The experimental results showed that the LAI binary linear regression measurement model established by Hr and Mr variables was the best. The R2 of Jingnongke 728 training set was 0.931, the verification set R2 was 0.949, the R2 of Nongda 84 training set was 0.979, and the verification set R2 was 0.984. The research result provided a solution for rapid measurement in the LAI field.
