UAV multi-source data fusion with super-resolution for accurate soybean leaf area index estimation.

Introduction: The Leaf Area Index (LAI) is a critical biophysical parameter for assessing crop canopy structure and health. Unmanned Aerial Vehicles (UAVs) equipped with multispectral sensors offer a high-throughput solution for LAI estimation, but flight altitude compromises between efficiency and image resolution, ultimately impacting accuracy. This study investigates the integration of super-resolution (SR) image reconstruction with multi-sensor data to enhance LAI estimation for soybeans across varying UAV flight altitudes. Methods: RGB and multispectral images were captured at four flight altitudes: 15 m, 30 m, 45 m, and 60 m. The acquired images were processed using several SR algorithms (SwinIR, Real-ESRGAN, SRCNN, and EDSR). Texture features were extracted from the RGB images, and LAI estimation models were developed using the XGBoost algorithm, testing data fusion strategies that included RGB-only, multispectral-only, and a combined RGB-multispectral approach. Results: (1) SR performance declined with increasing altitude, with SwinIR achieving superior image reconstruction quality (PSNR and SSIM) over other methods. (2) Texture features from RGB images showed strong sensitivity to LAI. The XGBoost model leveraging fused RGB and multispectral data achieved the highest accuracy (relative error: 4.16%), outperforming models using only RGB (5.25%) or only multispectral data (9.17%). (3) The application of SR techniques significantly improved model accuracy at 30 m and 45 m altitudes. At 30 m, models incorporating Real-ESRGAN and SwinIR achieved an average R2 of 0.86, while at 45 m, these methods yielded models with an average R2 of 0.77. Discussion: The results demonstrate that the fusion of SR-reconstructed imagery with multi-sensor data can effectively mitigate the negative impact of higher flight altitudes on LAI estimation accuracy. This approach provides a robust and efficient framework for UAV-based crop monitoring, enhancing data-driven decision-making in precision agriculture.