Partial System And Grading Software Of Quality Detection Of Jujube

Jujube has high nutritional value and economic efficiency. It plays a very important role in total fruit exports of China. However, the present grading measures of jujube are mainly artificial and mechanical. The artificial grading resulted in uneven quality and low production efficiency. Mechanical grading easily leaded to mechanical damage. Therefore, the purpose of this study is to gain rapid and non-destructive testing measure for jujube to ensure the grading quality.The study was focused on dry jujube. Machine vision technology and near-infrared(NIR) spectroscopy were applied to detect external and inside quality. The main research were as follows:(1) Finished the assembly of online detection device in jujube quality, including the conveying device, the machine vision system and the near infrared spectrum system. Machine vision system to achieve detection in the external quality of jujube, near infrared spectroscopy system to achieve detection in the internal quality of jujube.(2) The design of external quality grading software for jujube based on machine vision was completed, which includes four parts, the image data reading module, the setting module, the image processing module and the statistical results module, the image processing module include the size detection module, the disease detection module and the crack detection module. The gray, median filtering and binarization were used to pretreatment image of the jujube. The minimum external rectangle of the jujube was used to determine its size. The H value of HSI(Hue-Intensity-Saturation) color model was used to determine disease area of the jujube. The I value of HSI color model was used to determine the crack area of the jujube. The external quality testing and grading software was tested, and the disease parameters were adjusted by the experimental results, the recognition accuracy rate of the disease date reached 88.24 %, the recognition accuracy of the crack date reached 80.95 %.(3) Moisture and total sugar of the jujube applying the near infrared spectroscopy, the quantity analysis model was optimized. It showed that the determination coefficient and prediction accuracy of models were high. And design of the internal quality inspection and grading software based on near infrared spectroscopy was completed. The main function modules include spectral data reading module, effective spectral information extraction and date moisture and total sugar detection module, classification module and user interface module, etc. And design of the processing software based on near infrared spectroscopy was completed. The main function modules include loading spectroscopy, selecting calibration samples and prediction samples, spectral pretreatment, establishment of calibration model and prediction of prediction samples.