Search

Cinnamon is a valuable aromatic spice widely used in pharmaceutical and food industry. Commonly, two-cinnamon species are available in the market, Cinnamomum verum (true cinnamon), cropped only in Sri Lanka, and Cinnamomum cassia (false cinnamon), cropped in different geographical origins. Thus, this work aimed to develop classification models based on NIR-hyperspectral imaging (NIR-HSI) coupled to chemometrics to classify C. verum and C. cassia sticks. First, principal component analysis (PCA) was applied to explore hyperspectral images. Scores surface displayed the high similarity between species supported by comparable macronutrient concentration. PC3 allowed better class differentiation compared to PC1 and PC2, with loadings exhibiting peaks related to phenolics/aromatics compounds, such as coumarin (C. cassia) or catechin (C. verum). Partial least square discriminant analysis (PLS-DA) and Support vector machine (SVM) reached similar performance to classify samples according to origin, with error = 3.3 % and accuracy = 96.7 %. A permutation test with p < 0.05 validated PLS-DA predictions have real spectral data dependency, and they are not result of chance. Pixel-wise (approach A) and sample-wise (approach B, C and D) classification maps reached a correct classification rate (CCR) of 98.3 % for C. verum and 100 % for C. cassia. NIR-HSI supported by classification chemometrics tools can be used as reliable analytical method for cinnamon authentication., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2022 Elsevier B.V. All rights reserved.)

Large amount of information in hyperspectral images (HSI) generally makes their analysis (e.g., principal component analysis, PCA) time consuming and often requires a lot of random access memory (RAM) and high computing power. This is particularly problematic for analysis of large images, containing millions of pixels, which can be created by augmenting series of single images (e.g., in time series analysis). This tutorial explores how data reduction can be used to analyze time series hyperspectral images much faster without losing crucial analytical information. Two of the most common data reduction methods have been chosen from the recent research. The first one uses a simple randomization method called randomized sub-sampling PCA (RSPCA). The second implies a more robust randomization method based on local-rank approximations (rPCA). This manuscript exposes the major benefits and drawbacks of both methods with the spirit of being as didactical as possible for a reader. A comprehensive comparison is made considering the amount of information retained by the PCA models at different compression degrees and the performance time. Extrapolation is also made to the case where the effect of time and any other factor are to be studied simultaneously., (Copyright © 2022 The Author(s). Published by Elsevier B.V. All rights reserved.)