Parametric Surface Modelling for Tea Leaf Point Cloud Based on Non-Uniform Rational Basis Spline Technique.

Area of Science:

• * Botany and Plant Science
• * Computer Vision and Graphics
• * Computational Geometry

Background:

• * Plant leaf 3D architecture is dynamic, responding to growth and environmental factors.
• * While leaf point cloud acquisition and segmentation are advanced, 3D modeling techniques remain underdeveloped.
• * Accurate 3D leaf models are crucial for understanding plant development and stress responses.

Purpose of the Study:

• * To propose a parametric surface modeling method for accurate 3D fitting of tea leaf point clouds.
• * To develop methods for generating ordered and unordered point sets for fitting.
• * To evaluate the fitting accuracy and efficiency of the proposed NURBS surface modeling approach.

Main Methods:

• * Principal Component Analysis (PCA) for point cloud alignment.
• * Slicing point clouds and selecting sections to create a Point Set to be Fitted (PSF).
• * Fitting the PSF to a Non-Uniform Rational B-Spline (NURBS) surface, optimizing using Particle Swarm Optimization (PSO) to minimize fitting error (RMSE and MV).

Main Results:

• * The proposed method accurately fits simplified tea leaf point clouds.
• * Achieved fitting errors below 1 mm (RMSE) and 2 mm (MV) without PSO for simplified data.
• * Demonstrated the potential for modeling a wider range of leaves and incomplete point clouds in future work.

Conclusions:

• * A novel parametric surface modeling method effectively reconstructs 3D tea leaf geometry from point clouds.
• * The method is robust, achieving high accuracy even with significantly simplified point cloud data.
• * This work provides a foundation for advanced 3D plant modeling, with potential applications in phenotyping and stress analysis.