Screening the Effective Spectrum Features of Tobacco Leaf Based on GA and SVM

摘要

To improve the tobacco classification speed, it is necessary to shorten the data acquisition time and reduce the computational complexity of the hierarchy model. In this paper, we take the genetic algorithm to screen the tobacco spectrum characteristics, and set up the support vector machine (SVM) classification mode, then compared the feature selection recognition rate of 13 tobacco leaves grade before and after. The experiment results show that the recognition rate improves greatly after using genetic algorithm for feature selection, and reduce the data acquisition quantity. By using the genetic algorithm method, we can improve the classification speed of tobacco leaves grading on the premise of the correct classification rate.