Tobacco mellowing judgment method based on near infrared spectrum

The invention discloses a tobacco mellowing judgment method based on near infrared spectroscopy, which comprises the following steps: obtaining tobacco samples including training set samples and prediction set samples, and performing sensory evaluation on the tobacco samples to obtain sensory mellowing scores of the tobacco samples; carrying out sample pretreatment on the tobacco sample, and obtaining a near infrared spectrum of the tobacco sample; establishing a correction model between the sensory alcoholization score of the training set sample and the near infrared spectrum by adopting PLS; obtaining a predicted alcoholization score of the prediction set sample by using the correction model, and optimizing parameters of the correction model; and utilizing the optimized correction model to predict the alcoholization score of an unknown tobacco sample. By adopting the method disclosed by the invention, the sensory evaluation score of the tobacco leaf field can be quickly and effectively obtained to predict t