Premium
A high‐precision method evaluating color quality of Sichuan Dark Tea based on colorimeter combined with multi‐layer perceptron
Author(s) -
Zou Yao,
Ma Wanjun,
Tang Qian,
Xu Wei,
Tan Liqiang,
Han Deyang,
Tian Yun,
Yuan Yue
Publication year - 2020
Publication title -
journal of food process engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.507
H-Index - 45
eISSN - 1745-4530
pISSN - 0145-8876
DOI - 10.1111/jfpe.13444
Subject(s) - colorimeter , artificial intelligence , color difference , pattern recognition (psychology) , multilayer perceptron , color analysis , perceptron , mathematics , principal component analysis , computer science , artificial neural network , computer vision , optics , physics , enhanced data rates for gsm evolution
Instrumental examination of Sichuan Dark Tea (SDT) quality instead of human panel sensory evaluation is important for quality control. This study attempted to create a high‐precision method to rapidly and accurately evaluate SDT color quality. Colorimeter combined with multi‐layer perceptron (MLP) was utilized to extract CIELAB color parameters of dried tea, liquor, and infused tea, respectively, and established the prediction models of color attributes scores with the optimal color parameters selected by a principal component analysis (PCA). MLP models established could accurately predict color total scores of SDT, the glossiness of dried tea, and chroma of infused tea ( R p = 0.889–0.989; RMSEP = 0.393–0.631). Besides, models based on tea pigments could accurately predict infused tea color total scores ( R p = 0.920, RMSEP = 0.531). Parameter a * was significantly correlated with almost all of the color evaluation factors of SDT and seemed to be the characteristic color parameter. The color quality of Sichuan Dark Tea can be excellently estimated by the method utilizing colorimeter coupled with MLP. Practical applications To meet the requirement of dark tea production and the costumer's expectation, this study attempted to create a high‐precision method quickly and accurately evaluating color quality of SDT. Usually, in the massive production, tea color evaluated by eyesight is time‐consuming, subjective, and has poor accuracy due to visual fatigue of human. In this work, we optimized color parameters with PCA to improve the performance of MLP models established. The results can provide a theoretical basis for the evaluation of tea color quality by instrument, and make SDT quality control more convenient, accurate, and time‐saving.