Classification of tobacco leaf diseases based on multi-source remote sensing data

Accurate classification of tobacco leaf diseases is critical for objective disease assessment and management. However, traditional manual observation methods are inherently subjective, and classification approaches based on single-feature extraction often exhibit limited robustness. To address these limitations, this study proposes a tobacco leaf disease classification method based on multi-source data. Hyperspectral reflectance data, leaf area index, and chlorophyll content were selected as the original data sources, and corresponding feature extraction strategies were applied. Continuous wavelet transform was employed to extract discriminative features from hyperspectral reflectance data, while leaf area index and chlorophyll content were normalized using the Z-score method. A random forest algorithm was then used for model training and validation. Experimental results demonstrate that the proposed method achieves an overall classification accuracy of 88.7% with a Kappa coefficient of 0.83, indicating strong classification performance and robustness. These results confirm that the proposed multi-source data-based model provides a reliable and effective approach for tobacco leaf disease classification and offers valuable insights for future research using multi-source remote sensing data.