The growth quality of Pinus massoniana (Lamb.) seedlings is closely related to the survival rate of afforestation. Moisture content detection is an important indicator in the cultivation of forest seedlings because it can directly reflect the adaptability and growth potential of the seedlings to the soil environment. To improve the accuracy of quantitative analysis of moisture content in P. massoniana seedlings using near-infrared spectroscopy, a total of 100 P. massoniana seedlings were collected, and their near-infrared diffuse reflectance spectra were measured in the range of 2500 to 800 nm (12,000 to 4000 cm⁻¹). An integrated learning framework was introduced, and a quantitative detection model for moisture content in P. massoniana seedlings was established by combining preprocessing and feature wavelength selection methods in chemometrics. Our results showed that the information carried by the spectra after multiple scattering correction (MSC) preprocessing had a good response to the target attribute. The stacking learning model based on the full-band spectrum had a prediction coefficient of determination R² of 0.8819, and the prediction accuracy of moisture content in P. massoniana seedlings could be significantly improved compared to the single model. After variable selection, the spectrum processed by MSC and feature selection with uninformative variable elimination (UVE) showed good prediction effects in all models. Additionally, the prediction coefficient of determination R² of the support vector regression (SVR)—adaptive boosting (AdaBoost)—partial least squares regression (PLSR) + AdaBoost model reached 0.9430. This indicates that the quantitative analysis model of moisture content in P. massoniana seedlings established through preprocessing, feature selection, and stacking learning models can achieve high accuracy in predicting moisture content in P. massoniana seedlings. This model can provide a feasible technical reference for the precision cultivation of P. massoniana seedlings.

Pinus massoniana (Lamb.) is an important plantation species in southern China. Accurate measurement of P. massoniana seedling morphological indicators is crucial for accelerating seedling quality assessment. Machine vision, with its objectivity and stability, can replace human eyes in performing these measurements. In this paper, a measurement method for seedling morphological indicators based on Euclidean distance, Laplacian contraction, PointNet++, and 3D reconstruction is proposed. Firstly, multi-angle sequence images of 30 one-year-old P. massoniana seedlings were collected, distorted, and corrected to generate a sparse point cloud through the Structure-from-Motion (SFM) and dense point cloud through the Patch-Based Multiple View Stereo (PMVS). Secondly, a Dense Weighted Semantic Segmentation Model based on PointNet++ was designed, achieving effective segmentation of the P. massoniana seedling point clouds. Finally, a multi-iteration plane method based on Laplacian contraction was proposed. The new skeleton points were refined by minimizing the Euclidean distance, iteratively generating the optimal morphological skeleton, thus facilitating the extraction of morphological indicators. The experimental results demonstrated a good correlation between the machine vision-extracted morphological indicators (including plant height, ground diameter, and height-to-diameter ratio) and manually measured data. The improved PointNet++ model achieved an accuracy of 0.9448 on the training set. The accuracy and Mean Intersection over Union (MIoU) of the test set reached 0.9430 and 0.7872, respectively. These findings can provide reliable technical references for the accurate assessment of P. massoniana seedling quality and the promotion of digital forestry construction.