A Method for Estimating Plant Height of Facility Tomato Based on 3D Point Cloud

Plant height can be used to estimate biomass, indicate water stress, and can be also an effective indicator for nutrition content and yield. It is also an important parameter to assess crop growth in the field and to evaluate crop model performance. To achieve plant height for facility tomato quickly, accurately and non-destructively, a method was proposed to estimate tomato plant height based on image sequences. In this study, a truss equipped with visible light (RGB) camera system was used to obtain tomato canopy image sequences in the facility. Three-dimensional (3D) point cloud was reconstructed based on the structure from motion (SFM) algorithm. To estimate tomato plant height, the conditional Euclidean clustering algorithm was used to segment and extract tomato plants automatically from the 3D point cloud. The results showed that the root mean square error (RMSE) between estimation value and true value for tomato plant height was 4.62 cm and the coefficient of determination was 0.971. The method in this study can extract plant height for tomato effectively in a non-destructive way, which also provides a methodological guidance and platform for the other phenotypic parameters extraction for more facility crops.