Chao Zhao

Early detection of hidden maize weevil (Sitophilus zeamais) infestation in wheat is difficult, and moisture variation can obscure weak near-infrared spectral responses. This study investigated wheat in three nominal moisture groups designated as 13%, 16% and 19%. Hyperspectral data and corresponding three-channel pseudocolor images were acquired from healthy and infested kernels at four operationally defined sampling times to construct single- and mixed-moisture datasets. An Asymmetric Dual-Stream Masked Autoencoder Fusion Network (ADS-MAFNet) was developed by combining a CNN–Transformer spectral branch with masked autoencoder (MAE) pretraining and a 2D-CNN image branch for binary classification. Classification performance varied across the nominal moisture groups, with most spectral deep learning models performing less well in the 19% group. Within the investigated datasets, ADS-MAFNet achieved higher mean accuracy and F1-score than the evaluated single-branch neural models in the nominal 19% and mixed-moisture groups. Its mixed-moisture accuracy, precision, recall, and F1-score were 79.40%, 75.27%, 84.28%, and 79.44%, respectively. However, PLS-DA achieved higher mean accuracy and F1-score than ADS-MAFNet under all four evaluated moisture conditions. Ablation comparisons showed larger ten-run mean differences for branch removal than for removal of MAE pretraining, but a consistent advantage over IMG-only and a consistent pretraining benefit were not established. Classification performance was generally higher at the two later sampling times than at the two earlier sampling times. These results support further investigation of hyperspectral image fusion within the represented laboratory conditions; transfer to unseen moisture levels and independent batches remains untested.

The rice weevil (Sitophilus oryzae) is a major pest in stored wheat, and traditional detection methods face challenges in identifying its hidden life stages within kernels. This study develops a nondestructive method to detect S. oryzae (Sitophilus oryzae) infestation in wheat kernels using hyperspectral imaging, spectral preprocessing, feature extraction, and classification modeling. Hyperspectral data were collected from wheat kernels at different infestation stages (1, 11, 21, and 25 days (d)) and from healthy kernels. Spectral quality was optimized using SG smoothing, multiplicative scatter correction (MSC), and standard normal variate transformation (SNV). Feature extraction algorithms, including Competitive Adaptive Re-weighting Algorithm (CARS), Successive Projection Algorithm (SPA), and Iterative Retention of Information Variables (IRIV), were used to reduce data dimensionality, while classification models like Decision Tree (DT), K-nearest neighbors (KNN), and Support Vector Machine (SVM) were applied. The results show that MSC preprocessing provides the best performance among the models. After feature band selection, the MSC-CARS-SVM model achieved the highest accuracy for the 1 day and 25 d samples (95.48% and 96.61%, respectively). For the 11 d and 21 d samples, the MSC-IRIV-SPA-SVM model achieved the best performance with accuracies of 94.35% and 94.92%, respectively. This study demonstrates that MSC effectively reduces spectral noise and improves classification performance. After feature selection, the model shows significant improvements in both accuracy and stability. The study confirms the feasibility of using hyperspectral technology to identify healthy and S. oryzae-infested wheat kernels, providing theoretical support for early, nondestructive pest detection.