Hyperspectral imaging for discrimination of Keemun black tea quality categories: Multivariate calibration analysis and data fusion

• DOI: 10.1111/ijfs.14624
• Publication Date: 2020-05-12T19:20:25Z
• Authors: Guangxin Ren; Ying Liu; Jingming Ning; Zhengzhu Zhang
• Keywords: Data set; Black tea; Sensor Fusion

Summary Food fraud causes significant economic losses for the industry and generates distrust between consumers traders. Tea is one of most valued beverages throughout world, being vulnerable to economically motivated cheat. The objective study was develop potential hyperspectral imaging (HSI) allying multivariate analysis data fusion identify authenticity Keemun black tea quality categories. Data that integrated texture characteristics based on grey level co‐occurrence matrix visible near‐infrared spectral features via competitive adaptive reweighted sampling (CARS) as target modelling. Support vector machine (SVM) random forest (RF) were utilised classification samples seven grades. RF model using fused gave best performance with correct discriminant rate 92.70% prediction set. This demonstrated HSI coupled effective in identifying sample rank.