Abstract

This work reports that the measurements of the likeness or the uniqueness of the 1100–2500 nm reflectance spectra of intact canola seed determined from principal component analysis (PCA) approximated spectra with global H or neighbourhood H statistics were not associated with oil concentration within the seed. The absence of stability in association between the H measurements and oil content was related to inconsistency in the amount and distribution (between principal components) of the spectral variation correlated to the oil content within and between different batches of canola seed. PCA was used to approximate variation in the 1100–2500 nm, second order derivative, reflectance spectra of intact canola seed, acquired from 15 batches of seed samples. The first eight principal components (PCs) captured 97.14% to 99.35% of the total variance in the spectra. The amount of variation captured by individual components was independent of the number of samples in the batch and oil content within the seed. The pattern of variance distribution among principal components was inconsistent and highlighted the uniqueness of the origin of the spectral variation in each batch of canola seed. In this study, the strength of correlation between oil content and principal components was used as a measure of component significance to the analysis of oil in the intact canola seed. In the examined sets of spectra, oil content was correlated to the low order components, PC1 to PC4. In the 15 files of spectra, oil content showed the strongest correlation to PC2 in eight sets of data, to PC3 in four sets of data and to PC1 in three sets of data. The strength of association between oil and the individual components varied considerably in magnitude among examined files of spectra; r² = 0.28–0.81 for the first strongly correlated component, r² = 0.05–0.29 for the second and r² = 0.02–0.19 for the third. The position of the PCs in the correlation sequence was inconsistent and underlined differences of oil signal/spectral data interactions in the individual sets of data. Examination of principal component loadings showed that in the reported files of spectra, principal components correlated to the oil content frequently captured variance at segments, which denote absorptions specific and accidental to canola oil. The outline of the loadings did not conform to a single, regular pattern common to all sets of data. The reported results are in disagreement with rationale of the methodology, which involves the spectra matching techniques for validating the predictive efficiency of near infrared (NIR) calibrations. The reported results highlighted that the reliable NIR quantification of oil content from reflectance spectra of intact canola seed would require an independent validation for every acquired set of spectra.

© 2005 NIR Publications

Spatial mapping of proteoglycan content in articular cartilage using near-infrared (NIR) spectroscopy

Isaac O. Afara, Hayley Moody, Sanjleena Singh, Indira Prasadam, and Adekunle Oloyede
Biomed. Opt. Express 6(1) 144-154 (2015)

Rapid determination of the main components of corn based on near-infrared spectroscopy and a BiPLS-PCA-ELM model

Lili Xu, Jinming Liu, Chunqi Wang, Zhijiang Li, and Dongjie Zhang
Appl. Opt. 62(11) 2756-2765 (2023)

Comparison of a physical model and principal component analysis for the diagnosis of epithelial neoplasias in vivo using diffuse reflectance spectroscopy

Melissa C. Skala, Gregory M. Palmer, Kristin M. Vrotsos, Annette Gendron-Fitzpatrick, and Nirmala Ramanujam
Opt. Express 15(12) 7863-7875 (2007)

Quantitative analysis of bayberry juice acidity based on visible and near-infrared spectroscopy

Yongni Shao, Yong He, and Jingyuan Mao
Appl. Opt. 46(25) 6391-6396 (2007)

Rapid identification of the variety of maize seeds based on near-infrared spectroscopy coupled with locally linear embedding

Shu Liu, Zhengguang Chen, and Feng Jiao
Appl. Opt. 61(7) 1704-1710 (2022)