Analysis Of Near Infrared Spectroscopy Data Processing In Fruit Quality Detection

In recent years,with the Israeli Venture company Consumer Physics has developed a miniature spectrometer SCIo which has U disk size as daily use,just need connected with a mobile phone Bluetooth,after the scan you can check the composition of the measured substance,This shows the great potential of spectrum detection technology in food safety testing.Fruits high quality standards is whether the excessive pesticide residues,as well as the sugar content is appropriate.In this paper,we focus on the nondestructive detection of apple sugar,pesticide residues and variety identification based on near infrared spectroscopy.The main work of this paper includes: Near-infrared spectral data by MATLAB to be analyzed and pre-processed,Study of the spectral characteristics of Brix,pesticide and varieties those three qualities,then builds predictive model of this three parameters,to achieve spectral acquisition and quality parameter extraction.By comparison,differential direct derivative,S-G derivative,multiplicative scatter correction,wavelet transform those preprocessing algorithm are studied,Method for constructing the model of partial least square method and principal component analysis method and corresponding characteristic wavelength selection method "projection based on principal component analysis discriminant method" used in the characteristic wavelength selection is proposed,The experimental results show that the model can be used to model the spectral modeling of the full wave band and the characteristic wavelength spectrum,and it is proved that the projection of the main component has a significant effect on the selection of the characteristic wavelength.PLS-DA algorithm was used to identify the varieties of apple and citrus,and the recognition rate reached 100%.Finally,a fast processing and real time analysis scheme for cloud computing is proposed,design process and analysis software program to achieve real-time analysis and model optimization of fruit quality spectrum information.Original near infrared spectral data obtained in the detection process into the real-time analysis and service system,automatic calling matlab spectral preprocessing algorithm to preprocess the raw spectral data,subsequently matched NIR calibration model,complete the forecast of fruit quality parameters.By using this system,users can quickly get the data information of the content of each component,and can also store the matching model and the near infrared spectrum data in the database.Real time analysis and service system has the characteristics of the continuous updating of the spectrum database and the optimization of the model iteration.The methods and experiments in this paper are of great significance to the application of near infrared spectroscopy in the field of food safety detection,which can lay a solid foundation for the construction and development of future food safety monitoring cloud.