The Applications Of Tchebichef Image Moments In Analysis Of Near Infrared Spectra

Owing to the advantages of simplicity,rapid,non-destructive and eco-friendly,near infrared spectroscopy technique has been used in various fields.However,the wide and weak signal is the shortcoming of near infrared spectra(NIR)and is the root reason that near infrared spectroscopy technique didn't get the application for many years.It is a great challenge how to extract the useful information from the NIR spectra and to analyze the target component in the complex sample.In analysis of the NIR spectra,in order to build satisfying models,chemometrics method is playing an indispensable part to treating the signal of NIR spectra.The application of Tchebichef image moments based on NIR spectra was studied,the research work of this dissertation mainly includes.Chapter 1 The introduction mainly briefly introduces the development status of NIR spectra technique,the feature of obtained NIR spectra,the application of chemometrics in analysis of NIR spectra,the image moments and the application in analysis chemistry,and summarized briefly the main research contents of this dissertation are presented.Chapter 2 Applying Tchebichef image moment to extract the feature information of the target components in complex samples based on raw NIR spectra.The quantitative model was built by the proposed method that combined Tchebichef image moment with partial least square(TM-PLS).Compared with the results built by conventional partial least squares(PLS)and classical interval partial least squares(iPLS),TM-PLS is not only simpler,but also can effectively extract the important information and realize the accurately quantitative analysis of target components.The statistical parameters of built models and verification results indicated Tchebichef image moment is an effective pretreatment method and can provide more accurate and reliable analysis results,even though the interference of unknown samples,partial overlapping of target peaks appear in NIR spectra.Chapter 3 Simultaneous quantitative analysis of three components in mixture samples based on NIR spectra with temperature effect.The measurement of NIR spectra technique is sensitive to many external conditions change,especially,the variation of temperature limits the application range of the NIR spectra technology.Therefore,in this work,the influence of temperature on NIR spectra was studied.We can use Tchebichef image moment to extract the useful information of target components from the NIR spectra measured by different temperature,the quantitative model was established between the concentration of target components and the extracted feature information by stepwise regression.In this study,we made full use of the temperature effect to expand the information of sample,Tchebichef image moment can extract the useful information of target components,owing to their inherent invariance and powerful multi-resolution properties and obtain the satisfying quantitative model.Chapter 4 The common quantitative model for multiple NIR instruments.In practical application,more than one NIR instruments could be used to determinate the spectra of samples.The different instruments may result in greater prediction error when the established calibration model from one instrument(primary instrument)is applied directly to the spectra measured on other instruments.Therefore,a reliable multivariate calibration model is indispensable to one NIR instrument.In order to prevent the degradation in accuracy and reliability of the multivariate calibration model and avoid time-consuming full recalibration procedures,in this contribution,we proposed the common model in multi-instruments by Tchebichef image moment,owing to their inherent invariance and powerful resolution.This study indicated that the NIR spectra is different measured on different instrument,the extracted TMs with different orders was represented the feature information of target components and employed to establish the satisfying common quantitative models.Chapter 5 Includes main conclusions and outlook.