Implementation of an SfM-MVS-based photogrammetry approach for detailed 3D reconstruction of plants

In recent years, non-destructive and non-invasive methods for 3D plant reconstruction have gained importance in plant phenotyping. Morphological traits reflect a plant’s physiological status and serve as key indicators for precision agriculture, crop protection, and food quality assessment. Accurate and efficient 3D modelling enables objective, repeatable monitoring of plant development and health, supporting data-driven decision-making in agricultural and food research. This study presents a cost-effective and flexible photogrammetric methodology for analysing plant morphological traits under controlled laboratory conditions. The system includes an industrial RGB camera mounted on a robotic arm, a rotating platform with an adjustable plant holder, and stable illumination. The key steps involved camera calibration, exposure optimisation, fine-tuning of evaluation algorithm parameters (tweaks), setting the optimal camera-to-object distance, and reducing computational load for 3D model evaluation. Comparative testing revealed that the most effective calibration strategy integrated simultaneous calibration, pre-calibrated parameters, and adaptive fitting, ensuring high reconstruction accuracy and consistent model quality. The optimal acquisition parameters were a 50 milliseconds exposure time, a tweak value of 0.9, and a 16 cm camera-to-object distance. Using more camera positions with fewer frames per position proved more efficient than the reverse. The optimal configuration consisted of three height levels with 40 frames each. Automation and data reduction led to a 75% decrease in processing time, reducing the scan time from 8 minutes to 2.7 minutes per plant. The developed method proved to be a reliable, reproducible, and affordable tool for routine 3D analysis of plant morphology via close-range photogrammetry.