可见-近红外高光谱图像技术快速鉴别激光打印墨粉

发布时间：2018-05-11 19:22

  本文选题：高光谱图像 + 墨粉种类鉴别　； 参考：《发光学报》2017年05期

【摘要】：为了使用快速、无损的方法区分激光打印文件使用的墨粉种类,利用高光谱成像技术结合化学计量法对6种激光打印墨粉的光谱数据进行建模和种类鉴别的研究。利用可见-近红外高光谱成像仪采集400~1 000 nm波段内的光谱数据,采用Savitzky Golay平滑、标准化、多元散射校正和标准正态变量变换4种方法分别对光谱数据进行预处理,而后分别建立随机森林(RF)、K最近邻(KNN)、支持向量机(SVM)、偏最小二乘判别分析(PLS-DA)和簇类独立软模式(SIMCA)模型,进而实现激光打印墨粉的种类鉴别。利用准确率、拒识率和误识率3个指标作为模型评价标准。实验结果显示,SVM和PLS-DA模型的效果最佳,准确率为100%,拒识率和误识率为0。基于可见-近红外高光谱成像技术可以实现激光打印墨粉的快速种类鉴别。
[Abstract]:In order to use fast and lossless methods to distinguish the toner types used in laser printing documents, the spectral data of six laser printed toner were modeled and identified by hyperspectral imaging technique combined with chemometrics. The spectral data were collected by using a visible-near infrared hyperspectral imager in the wavelength of 400,000nm. The spectral data were preprocessed by Savitzky Golay smoothing, standardization, multivariate scattering correction and standard normal variable transformation. Then, the following models were established, such as KNN, SVMN, PLS-DAA and SIMCA, respectively, and the classification of laser printing toner was realized by using the model of random forest fission K nearest neighbor, support vector machine (SVM), partial least squares discriminant analysis (PLS-DA) and cluster independent soft mode (Simca), respectively. The accuracy rate, rejection rate and error rate are used as the evaluation criteria of the model. Experimental results show that SVM and PLS-DA model have the best effect, the accuracy is 100 and the rejection rate and false recognition rate are 0. Based on visible-near-infrared hyperspectral imaging technology, laser printing toner can be quickly identified.
【作者单位】： 文件检验鉴定公安部重点实验室(中国刑警学院);浙江警察学院刑事科学技术系;司法部司法鉴定科学技术研究所;
【基金】：文件检验鉴定公安部重点实验室(中国刑事警察学院)课题(2015KFKT09) 浙江警察学院校局合作项目(2016XJY014)资助~~
【分类号】：TP334.8;TP391.41
，

本文编号：1875256