| 刘津彤,张岚泽,姜红,陈相全,段斌,刘峰.化学通报,2022,85(2):259-263,246. |
| 差分拉曼光谱结合SVM对便签纸的鉴别分析 |
| Identification of note paper by Differential Raman spectroscopy based on support vector machine |
| 投稿时间：2021-07-19 修订日期：2021-08-06 |
| DOI： |
| 中文关键词: 差分拉曼光谱 支持向量机 便签纸 F检验 K折交叉验证 |
| 英文关键词:Differential Raman spectroscopy support vector machine note paper F-test K-fold cross-validation |
| 基金项目:中国人民公安大学2021年基科费重点项目(2021JKF212)、国家重点研发计划项目(2017YFC0822004)和南京简智仪器设备有限公司技术合作项目(20191218)资助 |
|
| 摘要点击次数: 983 |
| 全文下载次数: 0 |
| 中文摘要: |
| 基于差分拉曼光谱技术与支持向量机(Support Vector Machine, SVM)模型,提出了一种对便签纸类检材的快速可视化鉴别方法。实验获取了40组不同品牌便签纸样本的差分拉曼光谱数据,利用BP神经网络和差分技术完成谱图的除噪与基线校正后,借助F检验与主成分分析提取谱段信息,构建出支持向量机分类模型。实验结果表明,当设置Linear为SVM模型的核函数时,可以实现对样本测试集的完全准确划分,K折交叉验证的结果理想。相比于传统聚类分析手段,本方法可以在原始高维光谱数据中筛选出有效特征矩阵,且SVM模型兼具高效性和准确性,为公安实践中纸张类物证的区分鉴别提供一种新思路。 |
| 英文摘要: |
| Based on differential Raman spectroscopy and Support Vector Machine (SVM) model, a fast visual identification method for note paper materials was proposed. In this experiment, the differential Raman spectral data of 40 groups of different brand note paper samples were obtained. After denoising and baseline correction were completed by using BP neural network and differential technology, the effective spectral segment information was extracted by F test and principal component analysis, and the support vector machine model was built for classification. The experimental result shows that when Linear is set as the kernel function of SVM model, the sample test set can be divided completely and accurately, and the K-fold cross-validation results is ideal. Compared with the traditional clustering analysis method, this method can screen out the effective feature matrix from the original high-dimensional spectral data, and the SVM model has both efficiency and accuracy, which provides a new idea for the identification of paper evidence in public security practice. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|