Visible/near-infrared hyperspectral imaging combined with machine learning for identification of ten Dalbergia species.

Introduction: This study addresses the urgent need for non-destructive identification of commercially valuable Dalbergia species, which are threatened by illegal logging. Effective identification methods are crucial for ecological conservation, biodiversity preservation, and the regulation of the timber trade. Methods: We integrate Visible/Near-Infrared (Vis/NIR) Hyperspectral Imaging (HSI) with advanced machine learning techniques to enhance the precision and efficiency of wood species identification. Our methodology employs various modeling approaches, including Principal Component Analysis (PCA), Partial Least Squares Discriminant Analysis (PLS-DA), Support Vector Machine (SVM), and Convolutional Neural Networks (CNN). These models analyze spectral data across Vis (383-982 nm), NIR (982-2386 nm), and full spectral ranges (383 nm to 2386 nm). We also assess the impact of preprocessing techniques such as Standard Normal Variate (SNV), Savitzky-Golay (SG) smoothing, normalization, and Multiplicative Scatter Correction (MSC) on model performance. Results: With optimal preprocessing, both SVM and CNN models achieve 100% accuracy across NIR and full spectral ranges. The selection of an appropriate wavelength range is critical; utilizing the full spectrum captures a broader array of the wood's chemical and physical properties, significantly enhancing model accuracy and predictive power. Discussion: These findings underscore the effectiveness of Vis/NIR HSI in wood species identification. They also highlight the importance of precise wavelength selection and preprocessing techniques to maximize both accuracy and cost-efficiency. This research contributes substantially to ecological conservation and the regulation of the timber trade by providing a reliable, non-destructive method for identifying threatened wood species.

Identification of Guiboutia species by NIR-HSI spectroscopy.

Near infrared hyperspectral imaging (NIR-HSI) spectroscopy can be a rapid, precise, low-cost and non-destructive way for wood identification. In this study, samples of five Guiboutia species were analyzed by means of NIR-HSI. Partial least squares discriminant analysis (PLS-DA) and support vector machine (SVM) were used after different data treatment in order to improve the performance of models. Transverse, radial, and tangential section were analyzed separately to select the best sample section for wood identification. The results obtained demonstrated that NIR-HSI combined with successive projections algorithm (SPA) and SVM can achieve high prediction accuracy and low computing cost. Pre-processing methods of SNV and Normalize can increase the prediction accuracy slightly, however, high modelling accuracy can still be achieved by raw pre-processing. Both models for the classification of G. conjugate, G. ehie and G. demeusei perform nearly 100% accuracy. Prediction for G. coleosperma and G. tessmannii were more difficult when using PLS-DA model. It is evidently clear from the findings that the transverse section of wood is more suitable for wood identification. NIR-HSI spectroscopy technique has great potential for Guiboutia species analysis.

Improving the adaptability of the optical performance monitor by transfer learning.

Owing to their mighty fitting ability, the supervised learning-based deep-learning (DL) models have been widely used in the area of optical performance monitoring (OPM) to improve the optical monitors' performance. However, the supervised learning-based DL models used in OPM are based on two important premises. The first premise is enough training data with labels; the second premise is the same distribution of the training and test data. Nevertheless, it is hard to meet the two premises in the real-world environment where the optical performance monitors are deployed, since the data are unlabeled and the optical network environment is dynamic (such as component aging caused by slow parameter variation), causing the degradation of the monitoring performance. This is because the supervised learning-based DL models lack the adaptability of the dynamic environment. For the purpose of improving the optical performance monitors' adaptability, we propose a transfer-learning-based convolutional neural network model to maintain the monitoring performance in the dynamic optical network environment. The transfer-learning method can transfer the learned knowledge from the labeled data under an invariant optical network environment to the unlabeled data under a dynamic optical network environment. During the training phase, the maximum mean discrepancy (MMD) is applied to match the features extracted from the source and the target domains. When the trained model is deployed in the OPM monitor, the robustness of the system to the dynamic environment would be enhanced. Four signals (60/100 Gbps 16/64 QAM) under different working environments are used to verify the adaptability of the method. The influence of the MMD's weight rates, batch size, and weight parameters confirmed the effectiveness of our method.