Related Experiment Video
Updated: Aug 7, 2025
08:47
Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
Published on: February 9, 2024
1.5K
Improved weed mapping in corn fields by combining UAV-based spectral, textural, structural, and thermal measurements
Binyuan Xu1, Ran Meng1,2, Gengshen Chen3
1College of Resources and Environment, Huazhong Agricultural University, Wuhan, China.
Pest Management Science
|March 8, 2023
Summary
Thermal canopy temperature (CT) data significantly enhances weed mapping accuracy when combined with spectral, textural, and structural features. Integrating these multisource remote sensing measurements offers a novel approach for precision agriculture and improved crop production.
Area of Science:
- Agricultural Science
- Remote Sensing
- Data Fusion
Background:
- Accurate spatial-explicit weed mapping is crucial for effective weed management and minimizing crop yield losses.
- Unmanned aerial vehicle (UAV)-based remote sensing offers an efficient platform for timely weed mapping.
- While spectral, textural, and structural features are commonly used, thermal measurements like canopy temperature (CT) have been underexplored for weed mapping.
Purpose of the Study:
- To quantify the optimal combination of spectral, textural, structural, and CT measurements for weed mapping.
- To evaluate the performance of different machine-learning algorithms in integrating these features.
- To assess the added value of thermal measurements in improving weed mapping accuracy.
Main Methods:
- Utilized UAV-based remote sensing to collect spectral, textural, structural, and thermal (CT) data.
- Applied various machine-learning algorithms, including Support Vector Machine (SVM), Random Forest, and Naïve Bayes Classifier.
- Investigated data-fusion strategies combining different feature sets.
Main Results:
- Canopy temperature (CT) improved weed-mapping accuracies by up to 5% (Overall Accuracy) and 0.051 (Marco-F1) when used with spectral, textural, and structural features.
- The fusion of textural, structural, and thermal features yielded the highest performance (OA = 96.4%, Marco-F1 = 0.964).
- The SVM model demonstrated superior performance, outperforming Random Forest and Naïve Bayes Classifier.
Conclusions:
- Thermal measurements effectively complement other remote sensing data, enhancing weed mapping accuracy within a data-fusion framework.
- Integrating textural, structural, and thermal features provides the optimal strategy for weed mapping.
- This study introduces a novel, effective method for weed mapping using UAV-based multisource remote sensing, vital for precision agriculture and crop production assurance.
