Premium
Visible and Near‐Infrared Hyper‐Spectral Imaging for the Identification of the Type of Wax on Pears
Author(s) -
Li Baicheng,
Zhou Yao,
Zhao Mantong,
Hou Baolu,
Zhang Dawei,
Wang Qi,
Huang Yuanshen
Publication year - 2017
Publication title -
journal of food processing and preservation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.511
H-Index - 48
eISSN - 1745-4549
pISSN - 0145-8892
DOI - 10.1111/jfpp.12749
Subject(s) - wax , pear , calibration , hue , identification (biology) , near infrared spectroscopy , projection (relational algebra) , environmental science , mathematics , computer science , remote sensing , artificial intelligence , horticulture , materials science , optics , statistics , botany , geography , biology , physics , algorithm , composite material
This article presents a method for rapid, credible and noninvasive identification of the type of wax on pears using hyper‐spectral imaging (HSI) in the visible and near‐infrared (Vis–NIR) range (400–1,026 nm). The successive projection algorithm (SPA) was used to select the most effective wavelengths for wax type identification within a calibration set of 108 pears. This set was used to build identification models based on Multiple Linear Regression (MLR) and Linear Discrimination Analysis (LDA) using the relative reflectance values of the effective wavelengths. A prediction set of 72 pears was used to verify the reliability of the models and the results of both models were compared. SPA–LDA was found to be a better model than SPA–MLR, with an identification accuracy of 99.07% for calibration and 95.83% for the prediction sets. This demonstrates that Vis–NIR HSI is a potential candidate for wax type identification in a rapid, credible and noninvasive way. Practical Applications In the international markets and packing houses, many traders will wax the surface of pears to ensure they have a fresh appearance. Most traders use edible wax, but to reduce the cost, some unscrupulous traders may use industrial wax which is harmful to people's health. In recent years, a growing number of food products contain pears, such as Perry, sweetened pear pulp and dried pear. Therefore, it is crucial that the safety of the pear's surface can be ensured. In this study, the relative reflectance values of 11 effective wavelengths were selected by the Successive Projection Algorithm (SPA) as the independent variables of Linear Discrimination Analysis (LDA) based on hyper‐spectral imaging (HSI), and the obtained 11 effective wavelengths can be developed into an on‐line multi‐spectral system for food safety detection. This method allows nondestructive identification of the type of wax on pears and can automatically focus during the detection process.