Crop Leaf Phenotypic Parameter Measurement Based on the RKM-D Point Cloud Method.

Area of Science:

• Agricultural Science
• Computer Vision
• Biotechnology

Background:

• Accurate measurement of crop leaf phenotypic parameters (length, perimeter, area) is crucial for monitoring growth and estimating yield.
• Challenges in processing leaf point clouds, including noise and uncertainty, lead to inaccurate phenotypic parameter measurements.

Purpose of the Study:

• To propose and validate the RKM-D point cloud method for precise measurement of crop leaf phenotypic parameters.
• To address the limitations of existing methods in handling complex leaf point cloud data.

Main Methods:

• Developed the RKM-D method by integrating Random Sample Consensus with ground point removal (R), K-means clustering (K), Moving Least Squares (M), and Euclidean distance (D) algorithms.
• Captured point cloud data of pepper leaves across three growth stages (14, 28, and 42 days) using a stereo camera.

Main Results:

• The RKM-D method demonstrated high precision across all measured parameters.
• Leaf length: R² > 0.81, MAE < 3.50 mm, MRE < 5.93%, RMSE < 3.73 mm.
• Leaf perimeter: R² > 0.82, MAE < 7.30 mm, MRE < 4.50%, RMSE < 8.37 mm.
• Leaf area: R² > 0.97, MAE < 64.66 mm², MRE < 4.96%, RMSE < 73.06 mm².

Conclusions:

• The RKM-D point cloud method provides a robust and accurate solution for measuring crop leaf phenotypic parameters.
• This method enhances the reliability of crop monitoring and yield estimation through precise phenotypic data acquisition.