Open AccessStudy on Test method of Kiwifruit Hardness Based on Hyperspectral Technique
Author(s) -
Xinyuan Chen,
Lina Zheng,
Zhiliang Kang
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1453/1/012143
Subject(s) - hyperspectral imaging , principal component analysis , mathematics , pattern recognition (psychology) , kernel (algebra) , test set , artificial intelligence , biological system , statistics , computer science , combinatorics , biology
Flesh hardness is an important index to measure the quality and storage character of fresh fruit. At present, destructive test method is usually used to test kiwifruit hardness by sampling, while a new method based on hyperspectral technique is proposed in this study. Firstly, hyperspectral images of kiwifruit samples are collected in the spectral range of [400nm, 1000nm], then the hyperspectral images are pre-processed by standard normal transformation, and three sub-ranges of the best combination are selected through synergy interval Partial Least Square (siPLS). Secondly, kernel principal component analysis is adopted for dimension reduction of spectral bands of the best combination sub-intervals, and the first two principal components are input into the trained partial least square (PLS) as characteristic spectral bands. Finally, the test results are compared with physicochemical measurement values. The experimental results show that the Cross-validation Root Mean Square Errors (RMSEC) and the correlation coefficients of the training set and the testing set obtained by the proposed method are 0.2698/0.9315, and 0.3573/0.8738 respectively, indicating that it can effectively test kiwifruit hardness.