中国塑料 ›› 2021, Vol. 35 ›› Issue (9): 116-121.DOI: 10.19491/j.issn.1001-9278.2021.09.018

• 标准与测试 • 上一篇    下一篇

基于差分拉曼光谱对一次性塑料杯盖的分类研究

郭琦1(), 姜红1(), 吴克难2, 杨金颉1, 段斌3, 刘峰3   

  1. 1.中国人民公安大学侦查学院,北京 100038
    2.武汉理工大学计算机科学与技术学院,武汉 430070
    3.南京简智仪器设备有限公司,南京 210049
  • 收稿日期:2021-02-09 出版日期:2021-09-26 发布日期:2021-09-23
  • 作者简介:郭琦(2001—),女,本科生,从事刑事科学技术的研究工作,Gq593439826@qq.com
  • 基金资助:
    中国人民公安大学2021年度基科费重点项目(2021JKF212);国家重点研发计划项目(2017YFC0822004)

Classification of Disposable Plastic Cup Cover Based on Raman Difference Spectroscopy

GUO Qi1(), JIANG Hong1(), WU Kenan2, YANG Jinjie1, DUAN Bin3, LIU Feng3   

  1. 1.Institute of Criminal Investigation,People’s Public Security University of China,Beijing 100038,China
    2.Institute of Computer Science and Technology,Wuhan University of Technology,Wuhan 430070,China
    3.Nanjing Jianzhi Instrument Equipment Co,Ltd,Nanjing 210049,China
  • Received:2021-02-09 Online:2021-09-26 Published:2021-09-23
  • Contact: JIANG Hong E-mail:Gq593439826@qq.com;jiangh2001@163.com

摘要:

差分拉曼光谱是目前较为常用的检验方法之一,本文对计算机分类识别技术在案件现场常见一次性塑料杯盖的应用进行实验,获取了样本的差分拉曼光谱。根据谱图可将样本分为3类:聚丙烯(PP)、聚苯乙烯(PS)和聚对苯二甲酸乙二醇酯(PET),对比分析其中官能团峰位,再利用Calinski?Harabasz评价指标优选最佳聚类数为5类,进行K?均值聚类。随机选取40个样本建立支持向量机判别模型,提取谱图方向梯度直方图和灰度共生矩阵后合并成一个向量作为特征矩阵的一行,剩余8个样本作为测试集验证结果。结果表明,识别准确率达到到100 %。此高效识别方法具有检验成本低、分析速度快等优点,在公安机关实际办案过程中可起到技术参考作用。

关键词: 差分拉曼光谱图, 一次性塑料杯盖, Calinski?Harabasz评价指标, K?均值聚类, 支持向量机算法

Abstract:

Raman difference spectroscopy is one of the most commonly used testing methods. In this paper, a computer classification and recognition technology were applied for the test of a common disposable plastic cup cover at the scene of the case to obtain the differential Raman spectra of the specimens. According to the obtained spectrum, the specimens can be divided into three categories: PP, PS and pet. The peak positions of functional groups were analyzed and compared. The best clustering number was then determined to be 5 by using the calinski harabash evaluation index, followed by carrying out the K?means clustering. 40 specimens were randomly selected to establish a support vector machine discriminant model. The directional gradient histogram and gray level co?occurrence matrix were extracted and combined into a vector as a row of the feature matrix. The remaining 8 specimens were used as the test set to verify the results. The results indicated that the recognition accuracy reached 100 %. This efficient identification method has the advantages of low inspection cost and fast analysis speed, and it can play a technical reference role for the actual process of handling cases form the public security organs.

Key words: Raman difference spectroscopy, disposable plastic cup cover, calinski harabasz evaluation index, K?means clu?stering, support vector machine algorithm

中图分类号: