基于向量空间模型的岩屑LIBS光谱分类识别方法

发布时间：2018-01-01 08:02

  本文关键词：基于向量空间模型的岩屑LIBS光谱分类识别方法　出处：《光谱学与光谱分析》2017年09期 　论文类型：期刊论文

  更多相关文章： 激光诱导击穿光谱 向量空间模型 岩屑 分类识别

【摘要】：向量空间模型最初用于文献检索,该模型是通过对文献内容进行特征文本提取后,将文献转换到文本向量空间,然后在文本向量空间中通过计算文献的特征文本向量与检索文本的特征文本向量的相似度,实现文献的检索,该方法基于模式识别中模板匹配的最近邻原则。针对光谱数据的特点和模式识别中模板匹配的基本原则,将向量空间模型引入基于样品光谱的分类识别。通过训练集中光谱数据获得各样品的光谱数据模板,提取训练集中各样品光谱数据模板特征峰的波长和相对强度信息,构建特征峰信息数据库,计算获得特征峰信息权值,将光谱数据转换到特征峰向量空间,获得各样品光谱数据模板的特征峰向量,构建样品特征峰向量数据库。同理获得预测集样品光谱的特征峰向量,在特征峰向量空间中通过计算预测集样品特征峰向量与样品特征峰向量数据库中各样品模板特征峰向量的余弦值,完成对预测集样品的分类识别。以岩屑样品的LIBS光谱为研究对象,将向量空间模型应用于LIBS光谱的分类识别。分类结果表明,该方法能够实现对岩屑样品LIBS全谱的快速分类识别,且在对预测集光谱数据进行平均处理后,分类准确率为100%。提出的基于向量空间模型的LIBS光谱分类方法可以拓展应用于其他光谱数据的分类识别。
[Abstract]:Vector space model was used for document retrieval, this model is based on the literature content features of text extraction, document conversion to text vector space, and then through the similarity calculation of text vector literature and retrieval of text feature vector text in text vector space, the method of literature retrieval, based on the principle of the nearest neighbor template in pattern recognition, according to the basic principle of template matching. The characteristics of spectral data and pattern recognition in the vector space model is introduced, based on the classification of sample spectra. The spectral data obtained by the training set of spectral data template of each sample, extracting the training set of the spectra data template characteristic peak wavelength and relative intensity information. The construction of characteristic peak information database, obtained the characteristic peaks of weights, convert the spectral data to obtain the characteristic peaks of the vector space. The characteristic peaks of vector sample spectra data template, sample build characteristic peak vector database. This prediction set of characteristic peak vector sample spectra, characteristic peaks in the vector space by calculating the characteristic peaks of each sample prediction vector template sample vector and the sample characteristic peak characteristic peaks in the database vector cosine, complete classification of samples the prediction set. In LIBS spectra of debris samples as the research object, classification of vector space model is applied to the LIBS spectra. The classification results show that this method can achieve the full spectrum of debris samples LIBS rapid classification, and the average spectral data of the prediction set, the classification accuracy of classification and recognition of LIBS spectral classification method based on vector space model can be applied to other spectral data for 100%. is proposed.

【作者单位】： 中国海洋大学光学光电子实验室;
【基金】：国家自然科学基金项目(41503063,41106080)资助
【分类号】：TN249
【正文快照】： 引言激光诱导击穿光谱技术(laser induced breakdown spec-troscopy,LIBS)是一种高能量脉冲激光击穿样品产生瞬态等离子体的原子发射光谱技术,已经越来越多地渗透到各个研究和应用领域,如:环境监测[1]、生物组织分析[2-4]、材料成分在线监测[5]等。LIBS具有实时、原位、多元素，

本文编号：1363612