Oilseed rape is the second largest oil-bearing crop worldwide, and is planted on a global scale. Agricultural production is highly dependent on weather conditions; in regions with climate conditions suitable for oilseed rape production, soil can become waterlogged for different durations. Oilseed rape waterlogging can occur during various growth periods due to continuous rain, poor field drainage, or ponding at low-lying locations.

HSIs of oilseed rape leaves under different durations of waterlogging stress (0, 3, and 6 days) were collected to build three datasets (NY 22, NZ 19, and both combined). We extracted red–green–blue (RGB) images and visible and near-infrared (VNIR 400–1000 nm) spectra from a region of interest (ROI) in each HSI. Quadratic discriminant analysis (QDA), k-nearest neighbor (KNN), and support vector machine (SVM) classifiers were used to build classification models for comparing images and spectra of samples under different waterlogging levels among the three datasets, and to conduct training and prediction. From each dataset, 70% of the images were used for training, and the remaining 30% were used for testing. In the classification of full-wavelength HSIs (400–1000 nm), QDA and SVM exhibited high multivariate classification accuracy, reaching 77.37% and 95.90% accuracy, respectively. In contrast, KNN displayed low accuracy, but good identification and prediction ability for variety NZ 19. Six optimal wavebands of 529, 641, 698, 749, 856, and 979 nm were used as input for successive projections algorithm (SPA) classification and analysis. The QDA mode had better classification performance, with identification accuracies of 100% and 94.44%, respectively. Overall, the VNIR classification results exceeded those of image classification.

HSIs combine images and spectra to create high-resolution digital images. In agricultural research, HSIs can be used to simultaneously obtain graphic and spectral information to better distinguish differences between samples. We used the QDA, KNN, and SVM algorithms to classify HSIs and spectra of oilseed rape leaves under three levels of waterlogging stress, and found that the highest accuracies were for the QDA model and NZ19 dataset, followed by the SVM and KNN algorithms. The performance of the VNIR spectrum classification was better than that of image classification. To reduce modeling complexity, the SPA algorithm was applied to classify and analyze images and spectra at six optimal wavebands. The results show that the QDA model and NZ19 dataset always maintained high classification accuracy and stability; the corresponding images at the 698 and 979 nm wavebands showed high classification accuracy, and images at the 529 nm waveband had poor classification accuracy. Hyperspectral imaging technology is feasible and useful for the detection of oilseed rape waterlogging stress.

Fig. 1. PCA model of Score images at PC 1, PC 2, and PC 3.