Nondestructive detection of water content of jujubes based on visible-near infrared spectroscopy

Abstract. The chemistry component is not used as an evaluation indicator in jujube grading process. The paper focuses on detecting the water content of jujubes through the Visible and near-infrared spectroscopy. Visible and near-infrared spectroscopy is a fast and non-destructive detecting method. The water content is an important indicator of jujubes which can be used to evaluate the freshness of them. Prediction models can connect the water content of the jujube with its spectrum. In this way, the internal quality of jujubes can be easily detected as long as we get the spectra data of them.225 of 300 jujubes from Xinjiang were selected as a calibration set and the other 75 jujubes as a prediction set. Spectra preprocessed by standard normal variate (SNV), derivative and multiplicative scatter correction (MSC) and median filter (MF). And then, established partial least square (PLS) regression models. The conclusion showed that the correlation coefficient (Rp) could reach 0.8580 and root mean square of prediction (RMSEp) was 0.0306 under SNV-MF-PLS model. The residual predictive deviation (RPD) of the model was 1.84. The result convinced it is feasible to detect water content of jujubes based on Vis/NIR technology.