Maturity stage distinction of pear based on visible/near infrared spectroscopy technology

2nd Annual International Conference on Information System and Artificial Intelligence

物理学;计算机科学

Xu, Weiwei^1 ; Lin, Min^1 ; Huang, Yongmei^1 ; Liu, Huijun^1 ; Zhou, Xinqi^2

College of Metrology and Measurement Engineering, China Jiliang University, Hangzhou, China^1

Focused Photonics(Hangzhou) Inc., Hangzhou

310052, China^2

关键词: Multiplicative scatter correction;    Partial least squares discriminants;    Pretreatment methods;    Recognition rates;    Second derivatives;    Spectral pre treatments;    Visible/near infrared spectroscopy;    Visible/near-infrared spectroscopies;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/887/1/012030/pdf
DOI  :  10.1088/1742-6596/887/1/012030

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

This paper selected pear as the research object, established a model for distinguishing different maturity stages of pear which bases on visible/near-infrared spectroscopy, discussed the influence of four types of spectral pretreatment, such as the Savitzky-Golay (SG) smoothing, first derivative correction, second derivative correction, and multiplicative scatter correction(MSC) on the model, and verified that competitive adaptive reweighted sampling(CARS) algorithm can effectively extract the effective information from the spectrum. Maturity stage distinction model of pears was established based on Visible/Near Infrared Spectroscopy Technology by using partial least squares discriminant (PLSDA). It analysed four different commen pretreatment methods to improve the model effect. The result shows that the first derivative correction carries the best predicted accuracy, and the recognition rate is up to 86.11%.The information from the spectrum was filtrated from the pretreated spectrum by adopting CARS algorithm, to optimize the maturity stage distinction model of pears which was established before. The results demonstrates that the CARS algorithm can actually extract effective information, and the recognition rate is up to 97.22%.