摘要
利用衰减全反射傅里叶变换红外光谱（ATR—FTIR）技术采集猪、牛、羊革样品的红外光谱信息，分析皮革红外光谱特征。通过化学计量学对光谱数据进行处理，选择合适的光谱区域和预处理方法，建立定性鉴别皮革种类的判别分析模型（Ⅰ和Ⅱ）及方法。根据计算得到的马氏距离，由模型Ⅰ对样品“牛／羊”和“猪”革种类进行判别．然后由模型Ⅱ进一步对“牛”和“羊”革种类进行判别。通过验证集样品对模型性能进行评价，结果表明模型Ⅰ对猪革判别正确率为100％；由于牛、羊革的光谱信息极为相似，模型Ⅱ判别正确率有所降低。
The infrared spectra of pig, cow and sheep leather samples were collected by the attenuated total reflection-Fourier transfotan infrared spectroscopy （ATR-FTIR） technology. The characteristics of infrared spectra for the leathers were analyzed. The discriminant analysis models （ Ⅰ and Ⅱ） and methods for identification of leather classes were established via data processing using chemometric after selecting the appropriate spectra regions and pretreatment methods. Based on Mahalanobis distances which were calculated by the models, the ＂cow/sheep＂ and ＂pig＂ classes of leather samples were discriminated by model Ⅰ, and then the ＂cow＂ and ＂sheep＂ classes of leather samples were further discriminated by model Ⅱ. The performances of the models were evaluated with the validation samples. The results show that the discriminated accuracy of model I for pig leathers is 100%, while that of model Ⅱ is lower due to the extremely similar spectral information for cow and sheep leathers.
作者
林先凯
王宁
叶正茂
刘强
毛小慧
LIN Xiankai;WANG Ning;YE Zhengmao;LIU Qiang;MAO Xiaohui(Wenzhou Quality and Technical Supervision Testing Institute/National Center for Footwear Quality Supervision and Inspection （Wenzhou）, Wenzhou 325007, Chin)
出处
《中国皮革》
CAS
2018年第7期13-17,共5页
China Leather
基金
国家质检总局科技计划项目(2016QK170)
