Font Size: a A A

Study On Classification Method Of Flue Cured Tobacco Based On Spectrum And Chroma

Posted on:2017-04-28Degree:MasterType:Thesis
Country:ChinaCandidate:S C FangFull Text:PDF
GTID:2311330503971367Subject:Optics
Abstract/Summary:PDF Full Text Request
Flue-cured tobacco is one of the important economic crops in China. The first step is to objectively and reasonably classify the quality level of flue-cured tobacco in the processing. At present, the national standard 42 classifications(GB2635-92) are used in the market. It is according to some factors such as the tobacco part, identity,colour, maturity, leaf structure, oil content, residual wounds and so on. But the classification is depending on empirical methods such as the human eye view and sense of touch. The results would change due to factors such as personnel and emotional experience. It had a bigger error and losed objectivity, which was unfavorable for the effective use of flue-cured tobacco. With the development of science and technology, the emergence of near infrared spectrometer, high resolution digital camera and various new softwares provides a strong foundation of science and technology for the intelligence classification of flue-cured tobacco. Intelligent classification has the advantages of nondestructive testing, high efficiency, easy to operate and so on, so it is an inevitable trend to carry out intelligent and digital processing of flue-cured tobacco.The near infrared spectra of the flue cured tobacco leaves and their powders were collected by spectrometer, and the characteristics and differences of leaf spectra and powder spectra were discussed in this paper. It obtained five categories by clustering analysis using smoke powder information at 1470 nm, 1938 nm of significant peaks and significant trough at 2020 nm, 2216 nm. In addition to the fifth categories of these five categories, the reflectivity at four wavelengths increases with the increase of the category.The reflectivity of the fourth category was the larggest, the corresponding verification accuracy was also the highest, up to 84%. At the same time, the paper also used the digital camera to collect the color of flue-cured tobacco leaf, and the color istransformed into RGB color model,According to the rules of the R, G and B color components, five categories of flue-cured tobacco were summarized and analyzed.The range of R value of the first type flue-cured tobacco was in the range of 143-153,G value in 75-85, B value range in 8-16. The positive rate was 71.43%, and 20% of the flue-cured tobacco samples were found to be in the second category. The range of R value of second kinds of flue-cured tobacco was in the range of 146-164, G value range in 94-112, B value range in 9-21. The positive rate was 62.86%, and 17.14% of the samples were sentenced to fourth categories. The range of R value of third kinds of flue-cured tobacco was in the range of 152-169, G value in the range of 116-120, B value in the range of 30-35, and the positive rate was 80%. The fourth and the fifth class sentenced positive rates were 65.71% and 74.29%, respectively. Finally, by using biochemical component content of flue-cured tobacco(total nitrogen, nicotine,crude fiber, protein) of smoke powder to classification research, it was found that the classification accuracy was above 80%In above three kinds of flue-cured tobacco classification method, the chroma classification method is the most simple and has the strongest practicability. Spectral classification and biochemical component classification method have higher accuracy,but damage to the raw materials. Synthesis of various conditions, it can be found that the chrominance grading method is better.
Keywords/Search Tags:Flue-cured tobacco grade, NIR, chroma, cluster analysis
PDF Full Text Request
Related items