Accurate plant 3D reconstruction and phenotypic traits extraction via stereo imaging and multi-view point cloud alignment.

Introduction: Accurate 3D reconstruction is essential for plant phenotyping. However, point clouds generated directly by binocular cameras using single-shot mode often suffer from distortion, while self-occlusion among plant organs complicates complete data acquisition. Methods: To address these challenges, this study proposes and validates an integrated, two-phase plant 3D reconstruction workflow. In the first phase, we bypass the integrated depth estimation module on camera and instead apply Structure from Motion (SfM) and Multi-View Stereo (MVS) techniques to the captured high-resolution images. It produces high-fidelity, single-view point clouds, effectively avoiding distortion and drift. In the second phase, to overcome self-occlusion, we register point clouds from six viewpoints into a complete plant model. This process involves a rapid coarse alignment using a marker-based Self-Registration (SR) method, followed by fine alignment with the Iterative Closest Point (ICP) algorithm. Results: The workflow was validated on two Ilex species (Ilex verticillata and Ilex salicina). The results demonstrate the high accuracy and reliability of the workflow. Furthermore, key phenotypic parameters extracted from the models show a strong correlation with manual measurements, with coefficients of determination (R²) exceeding 0.92 for plant height and crown width, and ranging from 0.72 to 0.89 for leaf parameters. Discussion: These findings validate our workflow as an accurate, reliable, and accessible tool for quantitative 3D plant phenotyping.