Measurement of Maize Leaf Phenotypic Parameters Based on 3D Point Cloud

Area of Science:

• Agricultural Engineering
• Plant Phenotyping
• Remote Sensing

Background:

• Accurate measurement of plant height (PH), leaf width (LW), and leaf angle (LA) is crucial for maize growth monitoring, lodging resistance assessment, and yield prediction.
• Existing lidar-based methods for acquiring these parameters often suffer from low automation, long measurement times, and susceptibility to environmental interference.

Purpose of the Study:

• To develop and validate a novel, automated method for estimating maize PH, LW, and LA using lidar point cloud projection.
• To overcome the limitations of existing phenotyping techniques in terms of speed, automation, and environmental robustness.

Main Methods:

• Acquisition of 3D lidar point cloud data of maize plants during middle-late growth stages.
• Gaussian mixture model (GMM) for point cloud registration and enhancement, followed by noise filtering and stem-leaf segmentation using point cloud projection and Euclidean clustering.
• Determination of PH by vertical distance, LW via midvein fitting on projected contours, and LA from skeleton diagrams of plant structures.

Main Results:

• The proposed method achieved high accuracy in field validation: 99% for PH, 86% for LW, and 97% for LA.
• Demonstrated rapid and automated measurement capabilities during critical maize growth phases.

Conclusions:

• The point cloud projection-based method offers an efficient and accurate solution for automated maize phenotyping.
• This technology supports advancements in maize cultivation automation and precision agriculture.