Improving High-Throughput Phenotyping Using Fusion of Close-Range Hyperspectral Camera and Low-Cost Depth Sensor

Area of Science:

• Plant Science
• Remote Sensing
• Agricultural Technology

Background:

• Hyperspectral cameras are used in high-throughput phenotyping for early plant stress detection.
• Plant geometry and illumination complexly affect spectral data quality.
• Plant structure, leaf area, and inclination are critical for plant modeling.

Purpose of the Study:

• To develop a novel approach fusing hyperspectral imaging with 3D point clouds for improved plant phenotyping.
• To overcome limitations posed by plant geometry and illumination on spectral data.
• To enhance the precision of biological process detection in high-throughput phenotyping.

Main Methods:

• Fusion of a low-cost depth sensor and a close-range hyperspectral camera.
• Development of a new calibration and analysis method for combined sensor data.
• Application and validation in controlled soybean leaf experiments.

Main Results:

• Achieved 0.99 pixel resolution for the hyperspectral camera and 3.3 mm accuracy for the depth sensor.
• Demonstrated quantification and modeling of plant geometry and sensor configuration effects.
• Showcased the potential of 3D reflectance models to minimize geometry-related spectral effects.

Conclusions:

• The proposed fusion method provides more accurate spatial and spectral plant information.
• This approach significantly enhances the precision of biological process analysis in high-throughput phenotyping.
• The fused sensor system offers a promising tool for advanced plant phenotyping and stress monitoring.