Assessing the Performance of RGB-D Sensors for 3D Fruit Crop Canopy Characterization under Different Operating and...

Area of Science:

• Agricultural Engineering
• Robotics and Automation
• Computer Vision

Background:

• Precision agriculture leverages 3D sensing for optimized farm management.
• Low-cost RGB-Depth cameras enable wider adoption of 3D sensing in agriculture.
• RGB-D sensor performance is sensitive to environmental lighting conditions.

Purpose of the Study:

• To develop and apply a methodology for evaluating RGB-D sensor performance under varying lighting and distances.
• To assess the impact of geometrical and spectral features on sensor accuracy.
• To characterize the operational capabilities of RGB-D sensors in agricultural settings.

Main Methods:

• A novel methodology was designed to test RGB-D sensors under diverse ambient illuminance and distances.
• Evaluation included geometrical (resolution, precision) and spectral (color, NIR) data.
• The Microsoft Kinect v2 sensor was tested in an apple orchard environment.

Main Results:

• Sensor resolution and precision declined significantly at illuminance levels above 2000 lx.
• Closer measurement distances mitigated performance degradation under high illumination.
• Low illuminance (<50 lx) negatively impacted color data quality, suggesting a need for artificial lighting.

Conclusions:

• The developed methodology effectively characterizes RGB-D sensor performance across typical orchard lighting conditions.
• The Microsoft Kinect v2 demonstrated robust performance in outdoor environments, despite its indoor design.
• Understanding lighting and distance effects is crucial for successful implementation of RGB-D sensors in precision agriculture.