Maturity discrimination on fresh tobacco leaves based on skewed leaf color distribution patterns

In order to discriminate the maturity of fresh tobacco leaves intelligently, based on the extension of color information parameters of RGB (Red Green Blue) color model, discrimination models for fresh tobacco leaf maturity were developed by adopting different parameter systems and modeling methods. The results showed that: 1) There were significant differences among the leaf color skewness parameters of fresh leaves with different maturity levels, and different leaf color information parameters changed differently with the increase of maturity. 2) The fitting degrees of the discrimination models developed by BPNN (Back Propagation Neural Network) method were superior to those by a multiple stepwise regression method. Meanwhile, the adoption of the skewness parameters significantly improved the discrimination accuracy of BPNN models for the maturity of fresh leaves, among which the BPNN model based on the skewness parameters of leaf color had better integrative performance. Therefore, the BPNN model based on the skewness parameters of leaf color could be used to discriminate the maturity of fresh tobacco leaves.