激光与光电子学进展, 2018, 55 (6): 063003, 网络出版: 2018-09-11 复制并引用该论文  本文已被引 4 次  被引通知

图 & 表

图 1. 不同蜂蜜样本的中红外光谱

Fig. 1. Mid-infrared spectra of different honey samples

下载图片 查看原文

图 2. 当输入20维特征数据且应用SVM算法识别率为100%时测试集的实际分类和预测分类结果

Fig. 2. Actual and predicted classifications of test set using SVM algorithm when recognition rate is 100% and 20-dimensional feature data are input

下载图片 查看原文

图 3. 当输入为20维特征数据且应用SVM算法识别率为99.23%时测试集的实际分类和预测分类结果

Fig. 3. Actual and predicted classifications of test set using SVM algorithm when recognition rate is 99.23% and 20-dimensional feature data are input

下载图片 查看原文

图 4. 当输入为20维特征数据且应用LSSVM算法识别率为100%时测试集的实际分类和预测分类结果

Fig. 4. Actual and predicted classifications of test set using LSSVM algorithm when recognition rate is 100% and 20-dimensional feature data are input

下载图片 查看原文

图 5. 当输入为20维特征数据且应用LSSVM算法识别率为97.69%时测试集的实际分类和预测分类结果

Fig. 5. Actual and prediction classifications of test set using LSSVM algorithm when recognition rate is 97.69% and 20-dimensional feature data are input

下载图片 查看原文

图 6. 支持向量由1维投影到2维空间的变换图

Fig. 6. Transformation graph of support vector from 1-dimensional space to 2-dimensional space

下载图片 查看原文

表 1不同主成分累积方差贡献率

Table1. Cumulative variance contribution rate of different principal components

Dimension 1 2 3 5 10 15 20 Contribution rate /% 46.539 82.272 91.908 96.054 99.188 99.129 99.856

查看原文

表 2线性SVM和LSSVM分类器模型对不同维数特征数据的平均识别率

Table2. Average discrimination rate of different dimension feature data from linear SVM and LSSVM classifier models%

Method 5-dimension 10-dimension 15-dimension 20-dimension PCA-SVM 84.54 87.38 96.38 97.77 PCA-LSSVM 87.31 92.54 96.15 97.69

查看原文

徐天扬, 杨娟, 孙晓荣, 刘翠玲, 李熠, 周金慧, 陈兰珍. 中红外光谱法结合支持向量机快速鉴别蜂蜜品种[J]. 激光与光电子学进展, 2018, 55(6): 063003. Tianyang Xu, Juan Yang, Xiaorong Sun, Cuiling Liu, Yi Li, Jinhui Zhou, Lanzhen Chen. Mid-Infrared Spectroscopy Analysis Combined with Support Vector Machine for Rapid Discrimination of Botanical Origin of Honey[J]. Laser & Optoelectronics Progress, 2018, 55(6): 063003.