3D phenotypic measurement of bitter gourd seedlings based on monocular structured light

Accurate 3D morphological characterization is critical for precision breeding. However, traditional phenotyping methods suffer from inherent limitations such as low efficiency and loss of dimensional information. To overcome these limitations, this study proposes a robust three-dimensional phenotypic analysis framework based on the monocular Fringe Projection Profilometry (FPP) method. This method is specifically optimized for the high-precision measurement of young plant leaf features. The framework employs a hybrid decoding strategy combining 12-step phase shifting with complementary gray codes to resolve phase ambiguity and reconstruct dense point clouds at sub-millimeter resolution. It addresses the problems of leaf occlusion and adhesion using the section segmentation method. Based on the Oriented Bounding Box (OBB) method of principal component analysis (PCA) and Delaunay triangulation technology, it achieves precise quantification of key phenotypic parameters such as leaf length, width, inclination angle, and area. Systematic verification using 72 bitter gourd seedlings as samples indicates that this method has an extremely high consistency with manual measurement results: the determination coefficients (R²) for leaf length, leaf width, and leaf inclination angle reach 0.9992, 0.9991, and 0.9933 respectively, with corresponding root mean square errors (RMSE) as low as 0.68 mm, 0.54 mm, and 0.92°, and the R² for leaf area reaches 0.9996. This system achieves the complete process from raw scanning to parameter extraction in a low-cost, non-contact, and semi-automated manner, providing reliable data support for the digital perception and intelligent grading standardization of seedling growth dynamics in precision breeding.