首页 | 本学科首页   官方微博 | 高级检索  
     

常见皮革的近红外光谱快速判别分析
引用本文:周莹,徐晓春,陈启群,傅霞萍. 常见皮革的近红外光谱快速判别分析[J]. 光谱学与光谱分析, 2016, 36(9): 2802-2806. DOI: 10.3964/j.issn.1000-0593(2016)09-2802-05
作者姓名:周莹  徐晓春  陈启群  傅霞萍
作者单位:1. 浙江省检验检疫科学技术研究院, 浙江 杭州 311215
2. 浙江大学生物系统工程与食品科学学院, 浙江 杭州 310058
基金项目:国家自然科学基金项目(31401289)
摘    要:皮革种类的鉴定对生产控制、贸易过程和市场管理都具有重要意义,目前尚未有皮革种类鉴定的正式检测标准,皮革种类鉴定主要依靠感官法、燃烧法、化学溶解法、显微镜法等综合判定,对人员要求高,主观性强。利用近红外光谱技术对市场五种常见皮革样品(头层牛皮革、剖层移膜革、羊皮革、再生革和人造革)进行分析,利用判别分析法(discriminant analysis, DA)结合标准正态变量变换(multiplicative signal correction,MSC)、多元散射校正(standard normal variate,SNV)、一阶微分(first derivative)、二阶微分(second derivative)等光谱预处理方法进行分类鉴别。结果显示:上述五种革类在同一空间中重叠严重,但是利用样本反面光谱数据可以轻易将人造革和其他革类进行区分,误判率为1.2%,余下四种天然或加工革类重叠稍显严重,在同一个空间内同时对四种革类区分效果不是很理想,两两分组的分类效果较好,误判率13%~17.9%。而不同的数据处理方法在不同的判别分析模型中带来的效果也不尽相同,未发现一种能持续稳定为模型提供优化效果的预处理手段。上述数据说明采用近红外手段对于皮革种类进行判别是具有可行性的,由于本次取样来自于市场上的皮包、皮衣、皮带等最终产品,已经过染色、压花、覆膜等各种复杂处理,可能对模型带来一定的影响,如果能采取一些手段,如扩大样本量等,减弱这些影响,应该能得到更满意的结果。

关 键 词:近红外  光谱技术  皮革  判别分析  分类   
收稿时间:2015-06-15

Rapid Discriminantion of Common Leather Varieties by Near Infrared Spectroscopy
ZHOU Ying,XU Xiao-chun,CHEN Qi-qun,FU Xia-ping. Rapid Discriminantion of Common Leather Varieties by Near Infrared Spectroscopy[J]. Spectroscopy and Spectral Analysis, 2016, 36(9): 2802-2806. DOI: 10.3964/j.issn.1000-0593(2016)09-2802-05
Authors:ZHOU Ying  XU Xiao-chun  CHEN Qi-qun  FU Xia-ping
Affiliation:1. Zhejiang Academy of Science and Technology for Inspection and Quarantine, Hangzhou 311215, China2. College of Biosystems Engineering and Food Science, Zhejiang University, Hangzhou 310058, China
Abstract:Rapid classification of leather variety means important to product process control,trading process and market surveil-lance.There is no official detection standard on classification of leather variety for the present.By now the testers use organolep-tic method,burning method,chemical dissolution method,microscope method,or combination of them,to give a convincing re-sult.The testers are required to highly sufficiently experienced,and not influenced by subj ective factors.It also costs too much time.For the purpose of this research,spectra of five common varieties of leather samples (full-grain leather,split leather, sheep leather,reborn leather and manmade leather)were collected from market.Discriminant analysis combined with pre-pro-cessing method,including multiplicative signal correction (MSC),standard normal variate (SNV),first derivative and second derivative were used to classify the spectra above.It shows that the above five varieties of leather overlapped seriously in the same space.But manmade leather can be easily distinguished from the other four leather varieties using rear spectra,with the misclassified percent of 1.2%.The last four leather varieties covered each other partly in the same space,classify of any two of them can reach a lower misclassified percent,about 1.3%~1 7.9%.Different pre-processing method affected the discriminan-tion model positively or negatively with no regularity.None of these pre-processing methods was found to give a positive effect in a stable and persistent way.It can be concluded that it is feasible to discriminate the common leather varieties by near infrared Spectroscopy.All of the samples were taken from the finish products in the market (eg,handbag,belt,leather coat),which were processed in different ways (eg.tanning,knurling,dyeing).The different processes of the samples could bring an unfore-seeable influence to the model which may be reduced by some method,for example,increasing the number and variety of sam-ples.
Keywords:Near infrared  Spectroscopy,Leather  Discriminant analysis,Classification
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《光谱学与光谱分析》浏览原始摘要信息
点击此处可从《光谱学与光谱分析》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号