Non-Destructive Measurement of Three-Dimensional Plants Based on Point Cloud

Area of Science:

• Agricultural Science
• Computer Vision
• Plant Biology

Background:

• Accurate measurement of plant growth parameters is crucial for agricultural research and development.
• Non-destructive methods for quantifying plant characteristics, such as height and leaf dimensions, are challenging yet highly desirable.

Purpose of the Study:

• To develop and validate a methodology for three-dimensional (3D) reconstruction of growing plants using Kinect v2.0.
• To explore the measurement of plant growth parameters from 3D point cloud data.
• To enable non-destructive, quantitative analysis of plant morphology and development.

Main Methods:

• Preprocessing of 3D point cloud data including outlier filtering and surface smoothing for accurate plant registration.
• Segmentation of plant components (leaves and stem) using the locally convex connected patches method.
• Extraction of leaf feature boundary lines and calculation of leaf dimensions (length, width, area) and plant height using geometric algorithms and surface integrals.

Main Results:

• The proposed methodology effectively performs 3D reconstruction and registration of growing plants.
• Automatic extraction of plant information, including leaf and stem measurements, proved effective.
• Measurement accuracy for plant height and leaf parameters met required standards, demonstrating the system's utility.

Conclusions:

• The established 3D plant model derived from Kinect v2.0 data is key for comprehensive plant information studies.
• This method reduces inaccuracies caused by occlusion, providing a more realistic description of leaf shape.
• The approach is conducive to studying real-time plant growth status and development in agriculture.