Application of OSC in sugar content evaluation of chestnut based on Near Infrared Spectroscopy

摘要

It is important in chestnut industry to evaluate the sugar content of nuts since sugar content is one of parameters for classifying the fruit to different productions. Previous work had proved the near infrared (NIR) spectroscopy could be used to measuring the sugar content in intact and peeled chestnut nondestructively; however, the performance of the predictive model would need more improvement. In this work, the orthogonal signal correction (OSC) algorithm was employed to optimize the predictive models. The results shown that, for the peeled chestnut sample, OSC could increase the correlation coefficient (R2) of validation set from 0.8649 to 0.8961 while decrease the root mean and square error of prediction from 0.739 to 0.626. For the intact chestnut sample, this algorithm did not improve the model performance. The results indicated that the OSC had potential to optimized the prediction accuracy of sugar content in chestnut based on near infrared spectroscopy.