A novel mesh processing based technique for 3D plant analysis.

Area of Science:

• Plant biology
• Computational biology
• Agricultural science

Background:

• Automated, non-invasive plant phenotyping is crucial for plant biology and phenomics.
• Existing 2D image processing methods are simpler but less comprehensive than 3D mesh analysis.
• 3D mesh analysis offers potential for accurate, time-series morphological feature estimation.

Purpose of the Study:

• To present and validate a novel 3D mesh-based technique for temporal high-throughput plant phenomics.
• To assess the accuracy of the methodology for analyzing plant growth, specifically in Gossypium hirsutum (cotton).
• To demonstrate the broad applicability of the approach for quantitative 4D monitoring of plant phenotypic features.

Main Methods:

• Developed a methodology based on previously reconstructed 3D plant meshes.
• Implemented stages including morphological mesh segmentation, phenotypic parameter estimation, and plant organ tracking.
• Validated the technique using manual measurements of cotton plants over multiple time-points.

Main Results:

• Achieved mean absolute errors of 9.34% for stem height, 5.75% for leaf width, and 8.78% for leaf length.
• Obtained high correlation coefficients (0.88–0.96) between automated and manual measurements.
• Demonstrated 95% accuracy in temporal leaf matching with an average analysis time of 4.9 minutes per plant over four time-points.

Conclusions:

• The 3D mesh processing methodology is accurate and suitable for quantitative 4D plant phenotyping.
• The approach provides a robust alternative to traditional 2D methods for plant growth monitoring.
• This technique has the potential for wider application in plant science research.