1：Experimental Design and Method Selection:

An indoor UAV low-altitude remote sensing simulation platform was built to acquire thermal, multispectral, and RGB images before and after artificial inoculation with Sclerotinia sclerotiorum. Image registration and fusion methods based on SIFT were used to create a fused database. Temperature analysis and machine learning models (SVM, RF, KNN, NB) were applied for disease detection and classification.

2：Sample Selection and Data Sources:

Oilseed rape plants were grown in pots in a greenhouse, with 60 pots inoculated with Sclerotinia sclerotiorum. Images were acquired from Day 1 to Day 6 post-inoculation.

3：List of Experimental Equipment and Materials:

Equipment includes FLIR Tau2 thermal camera, XIMEA xiQ multispectral camera, Canon EOS 650D RGB camera, acquisition card, image transmission system, remote control, monitor, and an indoor UAV simulation platform.

4：Experimental Procedures and Operational Workflow:

Plants were inoculated, and images were captured using the simulation platform. Image preprocessing involved linear transformation for thermal images, reflectance correction for multispectral images, and normalization. Image registration and fusion were performed using SIFT and Hausdorff distance.

5：Data Analysis Methods:

Temperature differences were analyzed, and machine learning models were trained and tested using 5-fold cross-validation. Accuracy was calculated, and confusion matrices were used for evaluation.