Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.
Identification of whether Gastrodia elata Blume has been fumigated by sulfur based on Machine Vision Technology
Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org
Citation: 2021 ASABE Annual International Virtual Meeting 2100100.(doi:10.13031/aim.202100100)
Authors: Zimei Zhang, Zhian Zheng, Aichao Li, Shufeng Cheng, Wenjie Wang, Dalong Jiang
Keywords: bayesian; feature extraction; Gastrodia elata; identification of origin; image recognition; principal component analysis;
Abstract. Gastrodia elata Blume (G. elata) is a traditional and precious Chinese medicinal material. It has sedative, antioxidant, anticonvulsant, and other biological activities. It can be used to treat insomnia, depression, hypertension, and other diseases, and has high medicinal and edible value. However, it is one of the more serious species of sulfur fumigation. Although sulfur fumigation has the effect of brightening the appearance of G. elata and prolonging the storage period, it will lead to the change of its chemical composition and pharmacological action, and affect the quality of G. elata, which causes people's high attention. Therefore, it is of great significance to identify whether G. elata has been fumigated by sulfur. A common approach to identifying whether G. elata has been fumigated with sulfur or not is to detect the content of sulfur dioxide. However, this method requires laboratory conditions, tedious processing steps, and a long waiting time to get results. So it can not be timely evaluated of the quality of G. elata. What‘s more, there are lots of experienced people who directly identify it through it‘s appearance, but this method is too subjective. Therefore, in order to quickly and objectively identify whether G. elata is sulfur-fumigated, this study proposed a method based on machine vision technology. This method can make up for the deficiency in three aspects: damage to the medicinal materials, delayed results, and too strong subjectivity. First of all, this method collected 519 sulfur-fumigated and unsulfur-fumigated G. elata from three typical habitats to obtain high-quality images. After the preprocessing of image compression, denoising, and segmentation, color features (HSV color moments), shape features (length, width, width-length ratio, and area) and texture features (energy mean, energy standard deviation, entropy mean, entropy standard deviation, mean moment of inertia, the standard deviation of the moment of inertia, correlation mean and correlation standard deviation) were obtained. Secondly, we analyzed whether there was a significant difference between the indexes of sulfur-fumigated G. elata and unsulfur-fumigated G. elata in the three features. Then these three kinds of single features were combined into four kinds of compound features, namely, color-texture feature, color-shape feature, texture-shape feature, and color-texture-shape feature. In addition, the support vector machine discriminant model, naive Bayesian classifier, and BP neural network model were established to conduct the training and testing of samples based on single feature and compound feature respectively. The ratio of training sample number and test sample number is 3:1. Finally, the recognition accuracy and F1-score value of different recognition methods are compared and analyzed. Three conclusions can be drawn from the results of this experiment: Firstly, There were significant differences between sulfur-fumigated G. elata and unsulfur-fumigated G. elata in most indexes of the three features. Secondly, the recognition accuracy and F1-score value based on shape features were significantly lower than those based on other features. Furthermore, the recognition accuracy and F1-score value of SVM based on the compound features formed by color texture shapes were the highest, reaching 99.1% and 99.2% respectively. Therefore, the compound feature of color, texture, and shape was the best feature, and the support vector machine model was the best model. This study effectively identified whether G. elata had been fumigated with sulfur, and provided a theoretical basis for better evaluation of G. elata quality by machine vision technology, and a new possibility for more convenient monitoring of G. elata quality. It also contributed to the digitization, automation, informatization, and intelligentization of Chinese medicinal materials.
(Download PDF) (Export to EndNotes)