基于主成分分析和支持向量机的木材近红外光谱树种识别研究

发布时间：2018-05-15 00:06

本文选题：树种识别 + 主成分分析　；参考：《光谱学与光谱分析》2017年11期

【摘要】：为了探究一种新型高效的树种鉴别方法,以桉木、杉木、落叶松、马尾松和樟子松近红外光谱数据为研究对象,分别建立了基于主成分分析和支持向量机的木材树种定性识别模型。在主成分识别模型中,样本光谱数据经过预处理后绘制了其二维和三维主成分得分图,可以看出:主成分分析得分图能有效区分五种木材树种,且三维得分图比二维得分图更能直观、清晰展示树种之间的差异,表明主成分分析在可视化层面上可对小样本树种进行有效判别。在支持向量机识别模型中,分别建立了以遗传算法和粒子群算法为代表的智能算法优化支持向量机树种识别模型,结果显示,遗传算法-支持向量机模型的交叉验证最佳判别准确率为95.71%,测试集预测准确率为94.29%,算法用时134.08s;粒子群算法-支持向量机模型的交叉验证最佳判别准确率为94.29%,测试集预测准确率为100.00%,算法用时19.98s,表明基于智能算法支持向量机树种识别模型能够实现对木材树种的有效鉴别。该研究对近红外光谱分析技术在木材科学领域的应用进行了有益探索,为木材树种的快速识别提供了新方法。
[Abstract]:In order to explore a new and efficient method for tree species identification, the near infrared spectra of eucalyptus, Chinese fir, larch, Masson pine and Pinus sylvestris var. mongolica were studied. Based on principal component analysis (PCA) and support vector machine (SVM), the qualitative recognition models of wood species were established. In the principal component recognition model, the two and three dimensional principal component score maps were drawn after the sample spectral data were preprocessed. It can be seen that the principal component analysis score map can effectively distinguish five wood species. The three-dimensional score map is more intuitive than the two-dimensional score map, showing the difference between tree species clearly, indicating that the principal component analysis can effectively distinguish small sample tree species on the visual level. In support vector machine (SVM) recognition model, the intelligent algorithm, represented by genetic algorithm and particle swarm optimization algorithm, is established to optimize the tree species recognition model of SVM, and the results show that, The best discriminant accuracy of genetic algorithm-support vector machine model is 95.71, the accuracy of test set prediction is 94.29, the time of algorithm is 134.08s, and the best discriminant accuracy of particle swarm optimization-support vector machine model is 94.29. The accuracy of set prediction is 100.00s and the time of the algorithm is 19.98s, which indicates that the tree species recognition model based on intelligent support vector machine can effectively identify wood species. This study provides a useful approach to the application of near infrared spectroscopy in wood science and provides a new method for the rapid identification of wood species.
【作者单位】：北京林业大学数学系;中国林业科学研究院木材工业研究所;
【基金】：国家自然科学基金项目(31670564)资助
【分类号】：O657.33;S781.1

【相似文献】