Related Experiment Video
Updated: Aug 19, 2025
11:49
Using High Resolution Computed Tomography to Visualize the Three Dimensional Structure and Function of Plant Vasculature
Published on: April 5, 2013
21.2K
A graph-based approach for simultaneous semantic and instance segmentation of plant 3D point clouds.
Katia Mirande1,2, Christophe Godin1, Marie Tisserand1
1Laboratoire Reproduction et Développement des Plantes, Univ Lyon, ENS de Lyon, UCB Lyon 1, CNRS, INRAE, Inria, Lyon, France.
Frontiers in Plant Science
|November 28, 2022
Summary
This study introduces a novel graph-based method for precise plant organ segmentation from 3D point clouds, improving automatic plant phenotyping accuracy and structural consistency.
Area of Science:
- Computer Vision
- Plant Science
- Computational Geometry
Background:
- Automatic plant phenotyping relies on accurate 3D point cloud segmentation.
- Existing methods often lack global structural consistency assessment.
Purpose of the Study:
- To develop a robust, two-level graph-based approach for accurate plant organ segmentation.
- To ensure structural guarantees and improve consistency in plant phenotyping.
Main Methods:
- Utilizing local geometric and spectral features on a neighborhood graph for point classification.
- Employing a quotient graph for inter-organ relationships and consistency checks.
- Implementing a refinement loop to correct segmentation errors.
Main Results:
- Successfully segmented plant organs (stems, branches, leaves, apices) with high accuracy.
- Demonstrated improved global structural consistency compared to classical methods.
- Validated on synthetic and real 3D point cloud data of Chenopodium album and Solanum lycopersicum.
Conclusions:
- The proposed graph-based method offers fast, accurate, and structurally consistent plant organ segmentation.
- This approach advances automatic plant phenotyping capabilities.
- The method is effective for diverse plant species and data types.
